WT/DS135/RPage 194

IV. ARGUMENTS PRESENTED BY THIRD PARTIES

A. BRAZIL

1. Introduction

4.1 Brazil explains that the proceeding challenges the WTO-consistency ofFrance's 1 January 1997 Decree, which bans the manufacture, processing, sale and possession for sale,importation, exportation, domestic marketing, offer and transfer of all varieties of asbestos fibres andproducts containing them (the Decree or the ban).1 The ban has four narrow exceptions that applywhere no substitutes exist for chrysotile products. The substitute products that do exist generally aremore expensive than chrysotile products. Thus, the ban clearly operates to create a commercialadvantage for substitute products. According to Brazil, a ban is the most trade restrictive of measures.Therefore, the justification for any ban must be subject to the strictest scrutiny, especially as applied toa developing country such as Brazil. The ban has ended Brazilian exports of uncontaminatedchrysotile to France. In 1994 and 1995, France imported from Brazil 1,100 and 1,500 metric tonnes ofuncontaminated chrysotile, respectively. Since the ban took effect in 1997, France has not importedany chrysotile from Brazil.

4.2 According to Brazil, the importance of this proceeding extends far beyond the French ban –the proceeding is a test case. Will other WTO Members be allowed to ban products of developingcountries that can be safely used with appropriate, tested precautions simply to appease the public?Modern economies use hundreds of products that present health risks if they are misused, but thatpresent no risks if they are used properly. Uncontaminated chrysotile is one of them; if properly used,uncontaminated chrysotile presents no health risk. Similar products include organic fibres, man-madefibres, benzene, mercury, ammonia, nearly all forms of pesticide, etc. Societies regulate theseproducts to ensure they are used safely so as to protect the health of workers handling them directlyand of the general population which is exposed to them indirectly. The same treatment is appropriatefor uncontaminated chrysotile. Uncontaminated chrysotile—the only asbestos fibre Brazil mines andexports – is the safest by far of all asbestos fibres. In particular, it is much safer than amphibole, theasbestos responsible for current health problems from past exposure. All of the asbestos that Brazilmines, produces and exports is uncontaminated chrysotile. For this reason, Brazil’s chrysotileproducts are among the safest in the world. The medical explanation for these facts is set forth indetail in a recent bio-persistence study by Dr. David S. Bernstein, an expert in fibre toxicology(indeed, the EC often seeks his expertise on this topic).2

4.3 Brazil asserts that the primary issue in this proceeding is not - as the EC would suggest -whether asbestos can be hazardous to human health. It can. Years of misuse and unsafe utilization ofthe most hazardous form of asbestos – amphibole - have caused significant damage to health. Allcountries, including Brazil, regret the harm to human health caused by decades of exposure earlier thiscentury to amphibole produced and used worldwide. Brazil understands well the basis of the publicoutcry, experienced in many countries (including Brazil), that led the French Government tocommission the INSERM Report3 (a study focusing on the health effects of earlier, unsafe uses ofamphibole asbestos) and then to ban asbestos. France imposed the ban only one day after INSERMreleased its Report. The Report was commissioned and released to provide a scientific "cover" for apolitical decision that had already been taken. However, as a review of the INSERM Reportdemonstrates, the causes of asbestos-related health problems in France are past uses, especially in the

1Decree No. 96-1133, dated 24 December 1996, (J.O. dated 26 December 1996).2David S. Bernstein, Summary of the Final Reports on the Chrysotile Bio-Persistence Study (Geneva

Switzerland; 2 October 1998).3INSERM, Effets sur la santé des principaux types d’exposition à l’amiante, Les Éditions INSERM,

Paris, 1997 (INSERM Report).

WT/DS135/RPage 195

spraying of brittle amphibole on to fireproof buildings and, until quite recently, warships (flocking).Given the long latency period between exposure to amphibole and the onset of any related diseases,workers who were victims of heavy exposure with virtually no protection 30 years ago areexperiencing serious health problems today. The INSERM Report is based on analyses of theseworkers' health. The INSERM Report does not focus on data from studies of modern uses ofchrysotile. Moreover, in the Report, INSERM concedes that it was unable to produce "scientificallycertain" conclusions, but could present only an "aid to understanding" based on "plausible, thoughuncertain, estimates.4" Quite simply, the INSERM Report is an inadequate basis for the ban.

4.4 Brazil argues that it has a deep appreciation of the desire - indeed, the need - for theFrench Government to address public concern and protect public health. Brazil also understands thefrustration of being unable to remedy or even mitigate the health consequences of past exposure fromunsafe use of amphibole, and the frustration of being unable to take measures to remedy or decreaseexposure from flocked amphibole asbestos that is already in French buildings (because disturbingflocking increases exposure). However, when France approved the WTO Agreement, it agreed not torestrict trade merely to appease domestic sentiment, no matter how strong. Brazil cannot acceptFrance's adoption of a politically motivated measure that will neither (i) make those already sick fromasbestos exposure healthy; nor (ii) reduce risk to the healthy beyond existing levels of protectionguaranteed by modern, controlled uses of chrysotile. As the European Commission recently stated:

[V]arious national organisations, including the Health and Safety Executive in the United Kingdom,have made very disturbing projections about the numbers of deaths which are likely to be attributable toasbestos over the next few decades. However, it is important to note that these figures relate to pastexposures to mixed asbestos types, including the fibres which have already been banned. It would bewrong to use these statistics alone to justify a ban on the marketing and use of chrysotile because such aban would not lead to a lower risk of exposure for workers to asbestos which is already in place, norwould it reduce the number of deaths which are occurring today as a result of past exposure toasbestos.5

4.5 Modern uses of asbestos are or should be limited to chrysotile, which most parties, includingINSERM, agree is safer than other forms of asbestos. Moreover, modern uses are or should beconfined to products in which the fibres are bonded in a finished product and, thus, cannot escape,e.g., asbestos-cement products.6 For these and other reasons, modern uses are quite safe; they involve

4See para. 4.30 below.5Official Journal of the European Communities, C 135/108 (14 May 1999) (30 September 1998 answer

of Mr. Bangemann to Written Question E-2736/98 of Christine Oddy (PSE)). See also Official Journal of theEuropean Communities, C 13/123 (18 January 1999) (24 July 1998 answer of Mr. Bangemann toWritten Question E-1950/98 of Anita Pollack (PSE)) ("[I]t is important to mention that a new ban would notlead to a lower risk of exposure to existing asbestos for workers, nor would it reduce the number of deaths frompast exposure to asbestos. Possible contamination from asbestos in existing buildings (e.g. in relation tomaintenance activities and asbestos removal operations) will remain an important cause of exposure to workersfor many years.").

6Brazil notes that in the chrysotile-cement industry, the largest present-day use of chrysotile, themanufacturing process uses a water slurry mixture of chrysotile and cement. No dust or pollution is createdduring this process. See also American Lung Association, Asbestos, pp. 2 and 3(http://www.lungusa.org/air/envasbestos.html) ("Asbestos is rarely used alone, and it is generally safe whencombined with other materials with strong bonding agents. As long as the material remains bonded so thatfibers are not released, it poses no health risk."); National Cancer Institute, (1996), p. 3(http://www.ncih.nih.gov./clinpdq/risk/Questions_and_Answers_About_Asbestos_Exposure.html) ("Asbestosthat is bonded into finished products such as walls, tiles, and pipes poses no risk to health as long as it is notdamaged or disturbed (for example, by sawing or drilling) in such a way as to release fibers into the air . . ..[N]o fiber type can be considered harmless, and proper safety precautions should always be taken by peopleworking with asbestos.").

WT/DS135/RPage 196

exceedingly low levels of exposure (that often do not exceed even the "natural" levels in ambient air).Chrysotile is used in a very wide variety of products. It is used as a flame retardant, to strengthenfriction materials (e.g., truck brakes) and to create cement pipes for carrying water that are far lesssubject to corrosion, cracking and breaking than traditional cement pipes. In most applications,chrysotile is used because it increases public safety; thus, using other, less-efficient products in itsplace often decreases public safety. The use of chrysotile as a fire retardant needs no explanation.However, a discussion of its use in friction materials may be illuminating. Chrysotile is usedprimarily in truck brake pads, drum brakes and brake blocks to control heat build-up, thus maximizingfriction and stopping power. It is the preferred product for this application. As one of the authors ofan American Society of Mechanical Engineers (ASME) study commissioned by the EPA concluded:

(a) The "replacement/substitution of asbestos-based with non-asbestos brake linings will producegrave risks"; and

(b) "the expected increase of skid-related highway accidents and resultant traffic deaths wouldcertainly be expected to overshadow any potential health-related benefits of fibersubstitution."7

4.6 Brazil pleads that chrysotile's numerous public safety benefits - the many contributions itmakes to societies around the world - not be ignored in this proceeding, as they were when Francepassed its ban. In Brazil's view, the primary question in this proceeding is quite narrow - is acomplete ban necessary to protect public health or can public health be ensured by regulating modern,controlled uses of chrysotile and chrysotile products? The answer arrived at by those countries in theAmericas that have examined the issue closely, ranging south from Canada, to the United States, toBrazil, is that public health can be ensured by regulating modern controlled uses. France may, ofcourse, take measures that are designed to, and actually do, protect its citizens. However, the ban doesnot meet even this very generous characterization of the general rule set forth in the WTO Agreementon Technical Barriers to Trade (the TBT Agreement). France must not be allowed to impose a ban onimports and safe, modern uses of chrysotile as a response to public pressure. That the ban does notapply to man-made fibres produced in France, which the available scientific data show present greaterrisks when their use is not controlled and which have not been proven safer, confirms that the basis forthe ban may be political and economic, but is not scientific or medical.

4.7 Brazil argues that in many respects, the French reaction is identical to that of the United StatesEnvironmental Protection Agency (the EPA) promulgated in 1989, when it banned asbestos underpressure from panicked U.S. public opinion. The EPA was unable to justify its ban scientifically tothe United States Court of Appeals for the Fifth Circuit. After lengthy legal proceedings, the FifthCircuit ordered the EPA to reverse its decision and to acknowledge publicly that modern productscontaining chrysotile enclosed in a matrix of cement or resin do not pose any detectable risk to publichealth.8 (Today, although amphiboles are prohibited in the United States, a number of productscontaining non-brittle chrysotile are permitted, including the products manufactured by Brazil andpreviously manufactured in France from Brazilian chrysotile.) Unfortunately, France has adopted ameasure that unnecessarily and to no good effect impedes international trade.

4.8 Brazil makes the following claims regarding the ban: (1) the ban is inconsistent withArticle 2.2 of the TBT Agreement because it creates unnecessary obstacles to trade and is more traderestrictive than necessary; (2) the ban is inconsistent with Article XI of the General Agreement on In developing countries such as Brazil, the availability of low-cost, high-quality building and piping materials,such as chrysotile-cement products, is crucial. Substitute products are more expensive and thus less available tothose who need them most.

7See Corrosion Proof Fittings v. EPA, 947 F.2d 1201, 1224, n.25 (5th Cir. 1991) (Corrosion Proof)(written testimony of Mr. Arnold Anderson, ASME).

8Id.

WT/DS135/RPage 197

Tariffs and Trade 1994 (GATT 1994) because it is a quantitative restriction that is not excused by theexceptions in Article XI:2 or Article XX; (3) the ban is inconsistent with Article 2.8 of theTBT Agreement because it applies to asbestos but not to man-made fibres or other substitute productsand thus operates as a technical regulation setting forth an unnecessary design or descriptivecharacteristic; (4) the ban is inconsistent with Article 2.4 of the TBT Agreement because internationalstandards for producing and using chrysotile and chrysotile products exist and France should haveused them; (5) the ban is inconsistent with Article III:4 of the GATT 1994 and Article 2.1 of theTBT Agreement (national treatment) because it does not apply to domestic man-made fibres and othersubstitute products, which are like products to chrysotile; and (6) the ban is inconsistent withArticle I:1 of the GATT 1994 and Article 2.1 of the TBT Agreement (MFN) because, insofar as itbans imports of chrysotile and chrysotile products, but not imported like product substitutes, itimproperly discriminates among imports.

2. Factual Aspects

4.9 Brazil concurs with practically all aspects of Canada's presentation, agreeing (i) that theFrench ban was passed in response to public outcry in France over the deaths associated with theintensive exposure to amphibole that had taken place early on in the century; (ii) with thecircumstances and risks of exposure presented by Canada. In particular, it agrees with the statementthat exposure, even in asbestos product plants, has decreased significantly and that, apart from existingflocked amphibole, current exposure is limited, or could be limited, entirely to chrysotile; in contrast,past exposure and current exposure from past uses (e.g. flocking) included exposure to amphibole;(iii) that current levels of exposure to modern uses of chrysotile are not significant and are notassociated with substantial health risks; (iv) with the fact that current controlled-use policies andstandards which are accepted internationally are sufficient to ensure the health of chrysotile workersand others exposed to chrysotile and to guarantee their safety; and, (v) with Canada's argument thatthe INSERM Report has many defects and that it was not the reason for France's ban on modern,controlled uses of chrysotile and chrysotile products.

4.10 Brazil considers that a "battle of experts", with one side presenting experts who supportbanning chrysotile and the other presenting experts who oppose banning chrysotile would be, in thiscase, both uninformative and unnecessary because the INSERM Report and the Synthesis9, as a matterof law, not fact, cannot support the ban.10 This Report and the Synthesis have several defects thatrender them utterly incapable of supporting the ban.11 INSERM has not conducted original research,but merely based itself on existing studies and, furthermore, it has not examined all existing studies asit has deliberately excluded those that have established a distinction between chrysotile andamphiboles. More specifically, the shortcomings of the INSERM Report include the following. First,the Report completely fails to examine the modern uses of chrysotile and chrysotile products and,thus, ignores the current state of the industry. Instead, it focuses on the health effects of exposure toamphibole that took place in previous decades. INSERM concedes that it does not have "direct,certain scientific" data on the health risks associated with current levels of exposure to the modernuses of any form of asbestos, much less chrysotile.12 In short, INSERM does not examine current usesand exposure levels and does not distinguish among the different levels of risk associated with the

9INSERM, Rapport sur les effets sur la santé des principaux types d'exposition à l'amiante, Expertise

collective INSERM, Paris, 1997, (hereinafter "INSERM Report"); INSERM, Effets Sur la Sante des Fibres deSubstitution à l’Amiante-Synthèse, Expetise collective INSERM, Paris, 1998, (hereinafter "Synthesis").

10Brazil concurs with Canada that the weight of all available scientific evidence, including theINSERM Report, leads to the conclusion that the ban serves no purpose other than restricting trade.

11Brazil notes that, because it is only the INSERM Report which preceded the ban, the ban must besupported by the Report alone. Brazil has discussed both the Synthesis and the Report because the formerunderscores some of the defects of the latter.

12INSERM, (1998), Effets sur la sante des fibres de substitution à l’amiante-synthèse, Paris, p. 226.

WT/DS135/RPage 198

different types of asbestos fibres (chrysotile, the only type produced and exported by Brazil, as well asused in it, is accepted as being the safest of asbestos fibres, even by INSERM itself).13

4.11 Second, the INSERM Report fails to examine the efficiency of the ways in which workerexposure has been reduced through the use of air filters in mines and plants14, and employing masks,laundry services, etc. Third, it does not even compare the risks of the past to the risks associated withman-made fibres15 and substitute products (such as ductile iron or polyvinyl chloride (PVC) pipes).16

By the time INSERM began to examine substitutes, the ban had already been in effect for 1.5 yearsand, in any case, INSERM issued only a synthesis and not a complete report on these substitutes.INSERM concedes in its Report that it lacked the data required to recommend the banning ofchrysotile and only to allow its substitutes.17 INSERM emphasizes that because it is the structure (sizeand shape) of fibres that makes them toxic when inhaled, any substitute fibre must be viewed asdangerous to human health.18 Finally, INSERM concedes that, although the health data it applied tochrysotile are from past, massive and prolonged exposure to amphibole, the data being collected forsubstitutes is based on much lower levels of exposure, replicating modern conditions. Most telling isthat INSERM states that toxicity levels for "asbestos" as a whole (and not merely for chrysotile)would yield similar results to those obtained for substitutes if similar testing conditions had beenused.19

4.12 Brazil further argues that INSERM uses a linear risk model to assume illogically and withoutany evidence that a threshold does not exist for safe exposure.20 France and INSERM are forced tocommit this methodological error (the assumption) due to the fact that they had data from pastprolonged exposure to amphibole but not to current, much lower exposure to chrysotile.21 To justify

13Ibid., p. 409 ("France used asbestos much later and to a much lesser degree than other countries, anddoubtlessly the asbestos used contained a lower proportion of amphibole-type fibres. Because of thesedifferences, it is not possible to simply transpose to France the results of projections concerning mesothelioma[and cancer] cases prepared recently for Great Britain.").

14Brazil notes that the Cana Brava Mine, in Brazil, for example, has an exceedingly complex andeffective air filtration system. The mine is the first and only asbestos mine in the world to have been certified ascomplying with ISO 14001. It was certified by Det Norske Veritas of Rotterdam, the Netherlands.

15See, e.g., Cossette, M., Substitutes for Asbestos, 4 December 1998; Anderson, A., Fibres in FrictionMaterials, December 1998; Davis, J.M.G., The Biological Effects of Fibres Proposed as Substitutes forChrysotile Asbestos: Current State of Knowledge in 1998, 1998; INSERM Synthesis. Brazil notes that thesestudies demonstrate that substitute fibres, both when manufactured and used, are likely to present health riskssimilar to those from chrysotile.

16See Corrosion Proof, 947 F.2d pp. 1226-27 (even while banning asbestos, the EPA conceded thatductile iron pipes and PVC pipes present health (cancer) risks "similar" to those presented by asbestos-cementpipes).

17INSERM, Effets sur la sante des fibres de substitution à l’amiante-synthèse, Paris, 1998, pp. 376 and428. Brazil notes that the European Commission has also recognized this as an important issue: "There is a keyscientific issue which Member States and the Commission agree still needs to be clarified. This is an assessmentof the relative risk posed by the substitutes in comparison to the risk posed by chrysotile." Official Journal ofthe European Communities, C 13/35 (18 January 1999) (11 June 1998, Answer of Mr. Bangemann to WrittenQuestion P-1451/98 of Peter Skinner (PSE)).

18INSERM Synthesis, p. 2.19Ibid., p. 33.20According to Brazil, the assumption is contrary to logic because a threshold must exist given that

asbestos is ubiquitous in water and air. Only those who have suffered intensive, prolonged, exposure havecontracted asbestos-related diseases.

21See also Asbestos in Public and Commercial Buildings: A Literature Review and Synthesis ofCurrent Knowledge, Health Effects Institute - Asbestos Research (1991) at pp. 6-9, para. 6.2.2 (Health hazardscaused by asbestos at levels encountered in buildings today are "on the order of 50,000 times lower thanindustrial exposure levels of the past."); Report of the Royal Commission on Matters of Health and SafetyArising from the Use of Asbestos in Ontario (1984), Background Briefing Notes No.1 - "Health Effects ofAsbestos" (Current exposure of the general public is "thousands of times less" than occupational levels in the

WT/DS135/RPage 199

the ban on the modern uses of chrysotile, France/INSERM had to assume that significant risk ispresent at all levels of exposure, even at those that are insignificant, out of political self-interest.INSERM adopted the linear risk model despite the fact that studies cited by theEuropean Communities (hereinafter "EC") themselves indicate that "bricoleurs" are not at risk. Thestudy conducted by Iwatsubo et al.22 indicates that low, sporadic, intermittent and cumulativeexposure of up to 0.5 fibres/ml-years does not present increased risk of mesothelioma. In commentingupon the results of an earlier study, the authors note that "no significant risk was observed for thosewhose exposure was intermittent".

4.13 Brazil argues that a close examination of the INSERM Report reveals that: (i) prolongedexposure to amphibole (its past uses) is associated with severe health problems (a proposition withwhich everyone agrees); (ii) substitute fibres have similar structures and, thus, when subject toscientific scrutiny, are expected to have similar health effects at similar levels of exposure;(iii) insufficient data exists on the health effects of current levels of exposure to chrysotile andsubstitute fibres, but the available data suggests that their health effects would be the same; and(iv) the Report does not purport to be as conclusive as France would have all believe; rather, toovercome (iii) above, INSERM extrapolated from the data used in (i), which as it itself concedes"does not produce scientifically certain knowledge, but only an aid to understanding the implicationsfor risk management.23" Brazil contends that the ban has been based on the irrelevant data describedabove. France employs the linear risk model as a tool to make data on past uses relevant to theimposition of the ban. However, INSERM researchers themselves recognize the limitations of thismodel and clearly state that it cannot produce "scientifically certain knowledge," but can only serve asan "aid to understanding," based on "plausible, though uncertain, estimates.24" These "conclusions"do not support significant trade restrictions, much less the ban. Rather, they are merely a call forfurther research.

4.14 Brazil argues that recent research focusing on uncontaminated chrysotile demonstrates why itpresents no health risks whatsoever. According to Dr. David Bernstein's medical explanation25, theserpentine (braided) structure of chrysotile leads it to unravel in the lungs (whereas the tubularstructure of amphibole and substitute fibres does not allow them to unravel and is unchanging); andonce unravelled, the smaller and thinner particles are more easily and rapidly enveloped bymacrophages and/or expelled from the lungs. Dr. Bernstein's research demonstrates thatuncontaminated Brazilian chrysotile of less than 20 microns (the length that has been associated withpathogenicity for all fibres) is very quickly cleared. The clearance half-time is 1.3 days (and is2.4 days for fibres of a length of 5-20 microns). He concludes that, once in the lung, chrysotile fibresdefibrillate (or unravel), breaking down into shorter fibres. According to Dr. Bernstein, this result "isin stark contrast to amphibole asbestos where a portion of the fibres longer than 20 [microns] remainsindefinitely or with synthetic mineral fibres where even very soluble fibres are removed by dissolutionin the lung with half-times greater than this."26 He concludes that uncontaminated chrysotile's lack ofbio-persistence suggests that it has "little if any toxicological effect."27 However, it is of course a fact

three decades during and after World War II (p. 3); the best estimates are that exposure of current occupants ofasbestos-containing buildings "is 1,000 to 10,000 times lower than the average exposure of insulation workers inthe past" (Volume 2 p. 585).

22Iwatsubo Y. et al., Pleural Mesothelioma: Dose-Response Relation at Low Levels of AsbestosExposure in a French Population-based Case-Control Study, American Journal of Epidemiology, 1998,Vol. 148, N° 2.

23INSERM Report, pp. 239 and 414..24 Ibid., pp. 239 and 232.25Bernstein D., Summary of the Final Report on the Chrysotile Bio-Persistence Study,

Geneva, 2 October 1998 (document presented by Brazil to the Panel).26Ibid., p. 4.27Ibid., p. 10.

WT/DS135/RPage 200

that if used improperly, uncontaminated chrysotile could be dangerous, but that would be the case forvirtually all products in existence and not just chrysotile.

4.15 Brazil indicates that it mines, produces and exports only uncontaminated chrysotile andchrysotile products, and subjects mining, production and use to strict regulations. In 1990, it signedthe ILO Convention and Recommendation Concerning Safety in the Use of Asbestos (Convention 162and Recommendation 172). To ensure safety in the mining, manufacture and use of chrysotile andchrysotile products and to meet its ILO obligations, Brazil passed a primary law28 and decree29 onasbestos. In addition, the production and use of chrysotile and chrysotile products is governed by"national tripartite (government-industry-workers) agreements". These set exposure limits, andprocesses of production and safety procedures to be used to guarantee worker safety. Finally, theBrazilian Asbestos Association (ABRA), a watchdog organization comprised of asbestos producersand sellers, further regulates the safety of, and trade in, chrysotile and chrysotile products.

4.16 Brazil explains that the ILO Convention and Recommendation are international standards thatestablish safety procedures for the handling of chrysotile and chrysotile products. They follow theILO Code of Practice on Safety in the Use of Asbestos.30 The goal of the Code is to prevent the risksof exposure to asbestos and its harmful effects and to provide practical control procedures for its use.Convention 162 and Recommendation 172 recommend the controlled and safe use of asbestos. Theirwording clearly indicates that the replacement of asbestos fibres should only take place when it isestablished that this is necessary to protect worker health and when replacement is technicallyfeasible. The replacement of chrysotile asbestos fibres contained in modern materials or products (i.e.where it is sealed in a matrix and cannot be released into the environment) is not necessary since theseproducts do not pose any detectable health risks. International standards, such as Convention 162 andRecommendation 172, recommend the regulation of asbestos on the basis of the type of asbestos fibreemployed, the products in which certain fibres are included, and their planned use. Thus,Convention 162 and Recommendation 172 stipulate the prohibition of crocidolite and materialscontaining friable asbestos for flocking31, but permit many uses of chrysotile, including those centralto this dispute (asbestos-cement and friction products). They allow countries to prohibit other specificuses if national authorities deem this necessary for worker protection, but only on condition thatsubstitute products be subjected to a thorough scientific examination of their health effects.32

4.17 In 1995, Brazil passed Law No. 9055 to discipline the extraction, industrialization, use,commercialization and transportation of asbestos and of asbestos-containing products, as well as ofnatural and synthetic fibres of any source used for the same purpose. The Law (i) bans the processingand use of all types of asbestos, except chrysotile and chrysotile-containing products; (ii) bans thecrushing and spraying (flocking) of all types of asbestos, including chrysotile, and of all substitutefibres; (iii) provides the framework for the tripartite agreements in that it sets deadlines for thegovernment's confiscation of the operating licences of companies that do not execute the tripartiteagreements, establishes medical inspection requirements for workers, and sets exposure limits forthose who work with chrysotile and substitute fibres subject to annual reduction. (In compliance withArticle 2.4 of the TBT Agreement, the exposure limits are determined based, in part, on therecommendations of "international entities which are scientifically accredited"); (iv) prohibits miners

28Brazilian Law No. 9055 of 1 July 1995.29Brazilian Decree No. 2350 of 15 October 1997.30Safety in the Use of Asbestos, Code of Practice, International Labour Organization, Geneva, 1990.31Convention 162, Article 12.32Article 12 of Recommendation 172 states:(1) The competent authority, wherever necessary for the protection of the workers, should require the

replacement of asbestos by substitute materials, wherever possible.(2) Before being accepted for use in any process, all potential substitute materials should be thoroughly

evaluated for their possible harmful effects on health. The health of workers exposed to such materials shouldbe continuously supervised, if judged necessary. (Emphasis added.)

WT/DS135/RPage 201

and wholesalers from supplying chrysotile or substitute fibres to companies that do not comply withall provisions of the Law; (v) applies special restrictions to the use of chrysotile and substitutes inproducts currently considered to be the riskiest, such as textiles; (vi) calls for research into the healtheffects of chrysotile and substitute fibres and provides financing for the effort; and (vii) provides forprompt Department of Justice action against infractions.

4.18 Brazilian Decree No. 2350 implements the Law, and (i) requires that prior to marketing, allproducts containing chrysotile of imported or national origin bear a "seal of compliance to theBrazilian System of Certification", and provides for the development of the certification system;(ii) requires research into and confirmation of the health effects of chrysotile and its substitutes;(iii) establishes additional requirements for the tripartite agreements which apply to all mines andcompanies producing chrysotile and chrysotile products; (iv) establishes requirements for monitoringand controlling the use of chrysotile and its substitutes, and ensures that a record is kept of theexposure measurements made by companies while guaranteeing access to them; and (v) establishes apermanent National Commission on Amianthus (NCA) to ensure the safety of workers involved in thechrysotile or substitute fibre industry. The Decree also establishes certain bodies, such as the NCA,composed of government and industry officials as well as workers, to ensure worker safety.

4.19 The tripartite agreements (otherwise known as The National Agreements for the Furtheranceof the Safe Use of Asbestos) are required by both the Law and Decree. They are executed by theFederal Government of Brazil, the industries involved (e.g. the mining or asbestos-cement industry)and the workers in the industry (through their unions). They establish mandatory medical proceduresinspection and safety measures, as well as exposure limits. They also give workers certain rights, bothindividual and collective, within their industries. Their objective is to continuously work towardsimproved worker safety and to decrease exposure limits as well as actual exposure. First, tripartiteagreements set the maximum permissible exposure limits to 0.30f/cm3, with 50 per cent of allmeasurements being below 0.10 f/cm3 (and without there being any constant exposureabove 0.3 f/cm3, even when the workers exposed have special breathing equipment). Second, theyrequire the use of specific "collective protection" procedures to protect workers. The procedures areto include the installation of air filter and exhaust systems, the use of wet processes in the handling ofchrysotile (which reduces dust release and, thus, exposure), the sealing of workspaces and processes tolimit exposure, the demarcation of areas of exposure for warning, the prohibition of dry sandpaperingprocesses, the implementation of a daily programme for the washing, wetting or vacuum cleaning ofproduction sites, and provisions for a change of work clothes (which may not be taken off site), oflaundry services and showers for employees. Third, the agreements require employers to provideworkers with individual protection equipment that complies with relevant standards. Fourth, they alsorequire them to conduct regular and detailed environmental evaluations of working conditions as wellas to medically inspect their employees. All results are to be filed with the Control Commission onthe Safe Use of Asbestos and with the Brazilian Asbestos Association, known as ABRA. TheControl Commission is comprised of plant workers elected by their peers. Fifth, they require theprovision of worker education programmes to communicate the health risks of exposure to chrysotile,the measures which can be taken to reduce exposure, and the "multiplier effect" which tobaccosmoking has on exposure. Sixth, they make ABRA responsible for providing the companies withtechnical assistance regarding controls and preventive measures.

4.20 Founded in 1984, ABRA is an industry watchdog group composed of companies from Brazil'sasbestos industry. Its main goal is to oversee industrial activity in order to ensure that ABRAmembers comply with the Law, the Decree and the tripartite agreements, as well as to educateworkers, wholesalers and end-users of chrysotile asbestos and asbestos products on safe use. Toaccomplish this goal, ABRA has an extensive, independent, monitoring programme. Biannually, itconducts spot measurements at the facilities of its members. It maintains an ISO 9000-certifiedlaboratory and sends control samples once a year to independent laboratories in Edinburgh (AFRICA)and Paris (LHCF) to ensure the accuracy of its measurements. If the company that has been tested

WT/DS135/RPage 202

fails to meet the applicable exposure limits, ABRA sends it a letter and informs its suppliers. It thenprovides the company with a maximum number of days in which to comply, and instructs itssuppliers to withhold chrysotile and/or chrysotile products from the company until it is able to notifyits compliance. The agreement restates the requirements of both the Law and Decree, and developscertain safety procedures. In exchange for compliance (and dues), ABRA serves as a low-costrepository for state-of-the-art safe-use technologies, covering areas such as plant, air filter and processdesign. It attempts to encourage as well as facilitate safe use, with its overarching objective being toregulate the industry in such a manner as to render additional government regulation unnecessary.The regulatory regime (consisting of the Law, the tripartite agreements and ABRA itself), aligns theself-interests of the industry with those of its workers. The industry and the workers individually, aswell as through Safety Commissions and Unions, cooperate to reduce health risks. The result of thiscooperation has been the creation of an extremely safe workplace with very low exposure levels. Ingeneral, this system encourages individual plants to exceed applicable requirements in order toguarantee worker and user safety. At the Capivari Asbestos Cement Plant, which is the largestchrysotile-cement plant in South America, the on-site doctor did not report a single case of asbestos-related disease among the employees whose contact with asbestos had been limited to the plant.

4.21 With respect to regulation of asbestos in the United States, Brazil asserts that, in response topublic outcry based on sensationalist media reports on the dangers of asbestos, theEnvironmental Protection Agency (EPA) banned asbestos in 1989.33 It prohibited "at staged intervals,the future manufacture, importation, processing, and distribution in commerce of asbestos in almostall products […]". In reaction, a United States company that manufactured asbestos pipes, CorrosionProof Fittings, filed suit against EPA arguing that the ban was not based on scientific and medicalinformation. In a 1991 decision, the United States Court of Appeals for the Fifth Circuit called for thelifting of the ban and ordered EPA to issue new rules grounded in science.34 The Fifth Circuitconcluded that EPA had presented "insufficient evidence to justify its asbestos ban." 35 Specifically, itfound that the EPA had failed to (i) consider all of the necessary and relevant evidence, and (ii) "giveadequate weight to statutory language requiring it to promulgate the least burdensome, reasonableregulation" that would protect human health.36 Similarly, France has failed to (i) examine existingevidence on the modern, controlled uses of chrysotile, (ii) assess the danger associated with substituteproducts, and (iii) impose a regulation that is not more restrictive than necessary. In 1993, EPA liftedthe ban and issued new provisions regulating the production and use of asbestos and asbestosproducts.37 Based on a thorough scientific and medical review, EPA then authorized more asbestosproducts (18) than it banned (6). None of the uses that are banned are at issue in this proceeding. Ofthe authorized uses, two are central to Brazil's exports to France and had previously been allowed(they include chrysotile-cement products and chrysotile friction materials).38 Under existingregulations, the United States produced 6,890 metric tonnes of chrysotile andimported 20,900 metric tonnes in 1997.39 In the same year, it consumed nearly 21,000 metric tonnesof chrysotile, exported unmanufactured fibre for a total value of US$5,690,000 and manufacturedproducts for a total of US$197,000,000.40 Public health has not suffered in the United States andpublic outcry did not resume. United States regulations ban the dangerous uses of asbestos, andregulate those that are safe.

33EPA Final Rule, 54 Fed. Reg. 29460 (1989).34Corrosion Proof v. EPA, 947 F.2d 1201 (5th Circuit 1991).35Ibid., p. 1215.36Ibid.37EPA Final Rule, 58 Fed. Reg. 58964 (1993).38The current US regulations on this topic are set forth at 40 C.F.R. part 763, Sub-Part I (1998).39United States Government Geological Survey, Minerals Yearbook 1997, Volume I at 4-5.40Ibid.

WT/DS135/RPage 203

3. Legal Aspects

4.22 Brazil argued that, as it turns to Brazil's legal arguments, the Panel should recall Brazil'scomplex system of regulation that ensures public safety. The Panel should serve the same roleregarding France's political decision which the Fifth Circuit served regarding EPA's political decision- that of a neutral arbitrator. Brazil understands that the Corrosion Proof decision does not in the leastbind the Panel - the procedures, legal standards and status of the parties are quite distinct. However,the court there faced similar circumstances and issues, and in the face of contrary public sentiment,issued a very focused, well-reasoned opinion, which is precisely what Brazil seeks here.

(a) The Agreement on Technical Barriers to Trade

(i) Article 12 of the TBT Agreement

4.23 Brazil argues that a prohibition of the trade and use of a product, such as France's ban, is themost restrictive of all possible trade measures and must be closely scrutinized by the Panel. Itrequests the Panel to devote particular attention to the ban on imports from Brazil, which is adeveloping country (and from Zimbabwe, a least-developed country). In general, WTO agreementsprovide for the special and differential treatment of developing and least-developed country exports.In the context of the TBT Agreement, special provisions are set forth in Article 12, which obligesMembers that are developing technical regulations and standards to consider the special needs ofdeveloping and least-developed countries and to provide them with differential treatment. Article 12.2obliges France to "take into account the special development, financial and trade needs" of developingcountries and of least-developed countries, when developing its technical regulations. France did notmeet this obligation. Rather, it adopted an outright ban that advantages French producers of substitutefibres and products to the detriment of Brazil's chrysotile and chrysotile product producers (and toZimbabwe's detriment as well). Moreover, the ban has not contributed to improving public health inFrance.

4.24 France violated Article 12.3 which covers the "preparation and application" of technicalregulations and standards. Article 12.3 requires France to ensure that its technical regulations "do notcreate unnecessary obstacles to exports" from developing countries such as Brazil (and fromleast-developed countries such as Zimbabwe). However, France's ban applies to Brazilian (andZimbabwean) exports and creates, to say the least, an "obstacle" to their trade. The obstacle is"unnecessary" because it does not contribute to the supposed objective of increasing safety. The onlytrade permissible under the ban is that of chrysotile and chrysotile product substitutes. The risksassociated with substitute fibres are unknown, but they are suspect. Meanwhile the risks associatedwith the modern, controlled uses of chrysotile, are zero.

(ii) Article 2.2 of the TBT Agreement

4.25 Brazil argues that the ban is inconsistent with Article 2.2 of the TBT Agreement because it ismore trade restrictive than necessary to fulfil a legitimate objective. Once it is established that theDecree is a "technical regulation", the EC must demonstrate (and have the burden of proving) that fourdifferent conditions have been met if they are to argue that the ban is in fact consistent withArticle 2.2.41 To defend the ban, the EC must demonstrate to the satisfaction of the Panel that (i) theobjective of the ban is "legitimate", (ii) that it "fulfils" this legitimate objective, (iii) that it is not

41According to Brazil, general rules of pleading, but also Article 2.5 of the TBT Agreement confirm

that France has the burden of justifying its trade restrictive measure. According to Article 2.5, a standard shallbe "rebuttably presumed not to create an unnecessary obstacle to trade" when it pursues a legitimate objectiveand is "in accordance with relevant international standards." France cannot take advantage of this exception tonormal rules of pleading because, as demonstrated below, the ban is contrary to relevant international standards.

WT/DS135/RPage 204

"more trade restrictive than necessary" to fulfil the legitimate objective, and (iv) that France evaluatedthe health effects (i.e. "the risks non-fulfilment would create") on the basis of "available scientific andtechnical information". According to Brazil, the ban meets only the first of these four conditions.

4.26 Brazil argues that the Decree is a "technical regulation" within the meaning of theTBT Agreement. The ban sets out certain (i) product characteristics, (ii) process and productionmethods, (iii) administrative provisions, as well as (iv) packaging, marking and labelling requirementswith which compliance is mandatory. Article 1 of the Decree prohibits the production, importation,exportation, manufacture, transformation, sale and offer for sale of all types of asbestos fibres andasbestos-containing products (except those temporarily excepted from the ban by virtue of Article 2.I).Thus, the ban is explicitly directed at product characteristics (asbestos and asbestos-containingproducts) and at process and production methods (all forms of production, manufacture andtransformation of asbestos and asbestos-containing products). Both the prohibition imposed byArticle 1 and the procedures for implementing and reviewing the entitlement to the exceptions set outin Articles 2.II and 3 of the Decree are "applicable administrative provisions" relating to productcharacteristics and process and production methods. Article 4 of the Decree prescribes certainmarking and labelling requirements for those few asbestos-containing products excepted underArticle 2. Compliance with the ban is mandatory and violations are penalized under Article 5. Brazilargues that both France and the EC have conceded that the Decree is a technical regulation. In WTOdocument G/TBT/Notif.97.55, dated 21 February 1997, the French Government notified the ban to theTBT Committee as a technical regulation. Paragraph 3 of the Notification indicates that the ban wasbeing notified under Articles 2.9.2 and 2.10.1 of the TBT Agreement, both of which establishnotification obligations for technical regulations. The European Commission has also recognized thatthe ban is a technical regulation both in a 15 April 1997 document justifying the French ban andduring the 8 July 1998 consultations on this dispute. Therefore, both France and the EC concede thatthe ban falls within the scope of paragraph 1 of Annex 1 of the TBT Agreement and is a technicalregulation.

4.27 Brazil does not contest that the objective of protecting the health of French workers andconsumers is a "legitimate objective" within the meaning of Article 2.2 of the TBT Agreement.However, it argues that the ban imposed by the Decree creates an unnecessary obstacle to trade. Itdoes not in reality fulfil its stated objective, and is more trade-restrictive than necessary to protect thehealth of French workers and consumers. In using the word "fulfil" (as in the requirement that"technical regulations shall not be more trade-restrictive than necessary to fulfil a legitimateobjective"), the text of Article 2.2 requires the existence of a rational link between the regulation andits stated objective.42 However, this rational link is absent as the ban does nothing to accomplish itsobjective. It does not make those who are now sick healthy and removing it would not make any ofthose now healthy, sick. The lack of a rational link between the ban and its purported objective isdemonstrated by the following: (i) that asbestos-related health risks are due to old and alreadyprohibited uses of asbestos; (ii) that there are no detectable health risks associated with modern uses

42Brazil notes that, while no WTO panel or Appellate Body reports have addressed this issue under the

TBT Agreement, relevant precedents under the SPS Agreement exist: In Japan - Apples, the Appellate Bodyfound that an SPS measure was justified only if the Member imposing the measure demonstrated a "rationalrelationship" between the SPS measure and available scientific information. Japan - Measures AffectingAgricultural Products (22 February 1999), WT/DS76/AB/R, para. 84; similarly, in EC - Hormones, theAppellate Body required the EC to establish "an objective relationship between two elements, that is to say, anobjective situation that persists and is observable between an SPS measure and a risk assessment."EC - Measures Concerning Meat and Meat Products (Hormones) (16 January 1998), WT/DS26/AB/R,para. 189; the Appellate Body has also held that a finding that an SPS measure is not based on an actualassessment of health risks is "a strong indication" that the measure does not really protect health but is instead "atrade-restrictive measure in the guise of an SPS measure." Australia - Measures Affecting Importation ofSalmon (20 October 1998), WT/DS18/AB/R, para. 166. This is precisely the case with the ban.

WT/DS135/RPage 205

of chrysotile; and (iii) that health risks associated with substitute fibres remain unknown and aresuspect.

4.28 Brazil asserts that the health risks addressed in the INSERM Report are based on pastexposure to high levels of asbestos fibres (largely amphiboles) and to exposure to old uses of asbestos,such as flocking. In prohibiting future importation and sale of chrysotile and modern chrysotile-containing products, the ban does nothing to address the effects (today) of exposure between 1940 andthe early 1960s to extremely high levels of asbestos, mainly amphibole fibres. It does not cureworkers who now suffer because of long-term exposure in the past to amphibole, the use of which wasbanned in France in 1994, or to unregulated concentrations of fibres that are "50,000 times" higherthan the modern-day internationally recognized controlled-use level of 1 f/ml.43 Likewise, prohibitingthe future importation and sale of chrysotile and modern chrysotile-containing products does nothingto address the effects of exposure to (or the disturbance of) friable asbestos, mainly amphibole, inFrench buildings prior to the 1978 French ban on flocking. This was recognized byEuropean Commissioner, Mr. Bangemann, who, in response to a question posed by the EuropeanParliament, responded that "[I]t is important to mention that a new ban would not lead to a lower riskof exposure to existing asbestos for workers, nor would it reduce the number of deaths from pastexposure to asbestos."44

4.29 Brazil maintains that there are no detectable health risks associated with modern uses ofchrysotile. There is no rational link between the ban and its purported objective because modern usesof uncontaminated chrysotile are safe. Prior to the ban, more than 90 per cent of the chrysotileimported into France was used in the manufacture of chrysotile-cement products.45 Currently, thechrysotile is bound to the cement and encapsulated in it, without there being any loose or friablefibres. Furthermore, most chrysotile-cement products are produced in such a way that sawing ordrilling are unnecessary, and in the few instances when either or both are required, widely recognizedand well-established procedures have been developed for these tasks which prevent fibre release.46

Similarly, in all other modern uses of chrysotile, the fibres are sealed, bonded or encapsulated in theproduct. In no instance are loose, friable fibres allowed to be. Brazil contends that France does nothave credible evidence to suggest that (i) sealed, bonded or encapsulated chrysotile poses a health risk,(ii) concentrations of chrysotile fibres at or below the internationally-recognized controlled use levelof 1 f/ml present a health risk, and (iii) controls do not eradicate all risk throughout a product's life-cycle (from mining to manufacture, distribution, sale and use, and eventual disposal). On the otherhand, much science-based research concludes that the level of chrysotile encountered in the workplacetoday, or in buildings, presents no detectable health risk. After an exhaustive study of the existingscientific literature, the Health Effects Institute concluded in 1991 that the health hazards created byasbestos at the levels commonly encountered today are "unlikely to be large enough to be actually

43Health Effects Institute – Asbestos Research, Asbestos in Public and Commercial Buildings: A

Literature Review and Synthesis of Current Knowledge, Cambridge, 1991, pp. 6-9.44Official Journal of the European Communities, C 13/123 (18 January 1999) (24 July 1998 answer of

Mr. Bangemann to Written Question E-1950/98 of Anita Pollack (PSE)).45Le Déaut J.-Y. and Revol H., L’amiante dans l’environnement de l’homme: ses conséquences

et son avenir, Office parlementaire d'évaluation des choix scientifiques et technologiques,Assemblée nationale no. 329 / Sénat no. 41, 16 October 1997.

46ISO 7337, §§ 4 and 5 (pp. 2-9): Brazil notes that the cutting of plates or tiles for roofing is not asource of emission if ISO-7337 is followed. ISO-7337 addresses the use of chains to break pipes throughpressure, low-speed saws, saws equipped with a vacuum dust extractor, and, also, proper wetting of the materialsprior to any action. The cutting or grinding of all cement pipe (even that which does not contain chrysotile)emits silica in the air, in the absence of proper controls. The International Association for Research on Cancer(IARC) rates silica as a Type 1 carcinogen (for man), like asbestos. The worker who cuts any cement pipetherefore has an interest in following ISO-7337.

WT/DS135/RPage 206

observed and measured.47" This conclusion (which was reached by an independent United Stateshealth watchdog), confirmed the 1984 findings of the Ontario Royal Commission.48 Similarly, in thecase brought by Corrosion Proof Fittings against the EPA, the Fifth Circuit made the followingcomment on the risk of asbestos products relative to toothpicks:

"As the petitioners point out, the EPA regularly rejects, as unjustified, regulations that wouldsave more lives at less cost. For example, over the next 13 years, we can expect more than adozen deaths from ingested toothpicks - a death toll more than twice what the EPA predicts willflow from the quarter-billion-dollar bans of asbestos pipe, shingles, and roof coatings."49

Brazil concludes that, because there are no detectable risks attributable to modern uses of chrysotile,there is no rational link between the French ban and its purported objective.

4.30 Brazil asserts that the French ban induces consumers to use chrysotile substitutes, whosehealth risks are unknown, in place of chrysotile, whose risks are known. In his 1998 paper on thebiological effects of substitute fibres, Dr. J. M. G. Davis concluded that "replacement [of chrysotile bysubstitute fibres] is premature in the present state of our knowledge .... The need for full toxicologytesting of new fibre products is recommended before these products are marketed."50 This conclusionwas shared by the European Communities Directorate General for Consumer Protection Policies,which stated that "there is no significant epidemiology base to judge the human health risks [ofsubstitute fibres] … hence the conclusion that [the uses of] specific substitute materials pose asubstantially lower risk to human health, particularly public health, than the current use of chrysotile,is not well founded … ".51 The INSERM Report itself acknowledges that the risks associated withsubstitute fibres are unknown. INSERM "urgently" cautions against their use until further scientifictests are conducted. It states that "[T]he absence of epidemiological data concerning the long-termsafety of these substitute products should not obscure the results of experimental systems indicatingthe possibility that pathological modifications could result. It is urgently important that suitableresearch into this area be conducted prior to the widespread use of substitute fibres."52 Despite thisurgent warning from its own experts, the French Government banned chrysotile and not its substitutesthe day after it received the INSERM Report. Thus, the French Government knowingly shiftedconsumption from chrysotile used in modern ways, and for which there is no detectable health risk, tosubstitute fibres for which "experimental systems indicate the possibility that pathologicalmodifications could result". Brazil concludes, therefore, that the ban does not "fulfil" a legitimateobjective as required by Article 2.2 of the TBT Agreement. The rational link between the ban and itsstated health objective does not exist because, as demonstrated above, (i) asbestos-related health risksare due to old, already prohibited, uses of asbestos, and not to the modern uses of chrysotile, (ii) nodetectable health risks are associated with the modern uses of chrysotile, and (iii) substitute fibres,whose health risks are unknown, will replace chrysotile.

4.31 Brazil further argues that even if there were a rational link between the ban and the purportedobjective, the French ban would nonetheless be inconsistent with Article 2.2 of the TBT Agreement

47Health Effects Institute – Asbestos Research, Asbestos in Public and Commercial Buildings: A

Literature Review and Synthesis of Current Knowledge, Cambridge, 1991, pp. 6-9.48Report of the Royal Commission on Matters of Health and Safety Arising from the Use of Asbestos in

Ontario (1984), vol. 2, p. 585.49Corrosion Proof v. EPA, 947 F.2d 1201 (5th Circuit 1991. See also L. Budnick, Toothpick-Related

Injuries in the United States, 1979 Through 1982, 252 J. Am. Med. Ass’n., 10 Aug. 1984, p. 796 (which showsthat toothpick-related deaths average approximately one per year).

50Davis J.M.G., The Biological Effects of Fibres Proposed as Substitutes for Chrysotile Asbestos:Current State of Knowledge in 1998, p. 1 and 5.

51European Commission, DG XXIV, Opinion on a Study Commissioned by Directorate General III onRecent Assessments of Hazards and Risks Posed by Asbestos and Substitute Fibres (9 February 1998), p. 1.

52INSERM Report, p. 434.

WT/DS135/RPage 207

because it is "more trade-restrictive than necessary to fulfil a legitimate objective, taking account ofthe risks non-fulfilment would create.53" A ban is the most trade-restrictive measure possible. Itcould be justified only if France were able to prove that there was no reasonably available, less trade-restrictive, alternative. France cannot do so. Controlled use policies demonstrably fulfil the objectiveof protecting the health and safety of French workers and consumers. In assessing whether the ban ismore trade-restrictive than necessary, within the meaning of Article 2.2, the Panel should examineboth the risks of non-fulfilment and whether a less trade-restrictive measure is available to fulfil theobjective.

4.32 Brazil argues that available scientific and technical information does not support theimposition of the ban. Article 2.2 provides that, in assessing the risk that a technical regulation ismeant to address, Panels should consider, inter alia, relevant scientific and technical information,related processing technology and intended end uses of products. The risk to be avoided in the disputeat hand is the risk of illness resulting from exposure to (a) modern uses of chrysotile and chrysotile-containing products and (b) the disturbance of previously installed friable asbestos (largelyamphibole) in buildings. Illnesses associated with old, previously banned, uses of asbestos are notrelevant to this analysis. Moreover, they cannot be addressed through the present ban on trade,domestic sale and use. The INSERM Report, which provides the supposed scientific justification forthe ban, does not assess the health effects of current levels of exposure to modern uses of chrysotile.To determine the health risk associated with exposure to low levels of bonded, sealed andencapsulated chrysotile in chrysotile-cement and other modern applications, it applies the same risk ofexposure associated in previous decades with higher levels of exposure to friable asbestos (largelyamphibole). There is no scientific logic for such an extrapolation. The INSERM Report itselfconcedes that its conclusions are not "scientifically certain" but are merely "plausible, thoughuncertain, estimates". Several other scientific reports concur that there is no detectable health riskfrom modern uses of chrysotile.54

4.33 Brazil explains that all modern-day uses of chrysotile involve bonding, sealing orencapsulation. Such uses, or modern products, do not contain loose, friable chrysotile fibres – whichwere the cause of past asbestos-related illnesses. The risk associated with modern use is undetectable.Most modern products are manufactured to specifications well known in the building and publicworks trades, so that sawing or drilling operations are seldom necessary. When sawing or drilling arenecessary, there are well-established procedures to ensure that workers are not exposed to fibrerelease. Thus, Brazil argues that neither available scientific information, intended end-uses norprocessing technology, necessitate a ban on chrysotile. Brazil argues that while the term "necessary"has not yet been interpreted in the context of Article 2.2 of the TBT Agreement, the interpretationprovided by the Panel in the case on Section 337 of the Tariff Act of 1930 is instructive:

"[I]t was clear to the Panel that a contracting party cannot justify a measure inconsistent withanother GATT provision as 'necessary' in terms of Article XX:(d) if an alternative measurewhich it could reasonably be expected to employ and which is not inconsistent with otherGATT provisions is available to it. By the same token, in cases where a measure consistent withother GATT provisions is not reasonably available, a contracting party is bound to use, amongthe measures reasonably available to it, that which entails the least degree of inconsistency withother GATT provisions."55

53Brazil notes that, similarly, under the SPS Agreement, in determining whether an SPS measure is

more trade-restrictive than required, the authorities must evaluate whether an alternative, less trade-restrictive,SPS measure would achieve the importing country’s appropriate level of protection. See Australia - MeasuresAffecting Importation of Salmon (20 October 1998), WT/DS18/AB/R, paras. 208-210.

54See paragraph 4.29 above, regarding the conclusions of the American Health Effects Institute, theRoyal Commission and the U. S. Court of Appeals for the Fifth Circuit.

55United States – Section 337 of the Tariff Act of 1930, adopted on 7 November 1989, BISD 36S/345,pp. 392-93, para. 5.26.

WT/DS135/RPage 208

4.34 Brazil argues that the focus is on the range of measures "reasonably available" to France. Justas under Article XX(d), under Article 2.2 of the TBT Agreement, the French ban cannot be justified as"necessary" since a less trade-restrictive measure that fulfils the legitimate objective is available.There are numerous examples of controlled use that are both readily available and effective inaddressing the health risks associated with the modern uses of chrysotile. First, the ILO Conventionand Recommendation Concerning Safety in the Use of Asbestos (Convention 162 andRecommendation 172) establish procedures to ensure safety in the handling of chrysotile andchrysotile products. Second, the ILO's 1990 Code of Practice on Safety in the Use of Asbestos detailsappropriate controlled use procedures to ensure worker safety with respect to all chrysotile-containingproducts currently in use. Third, Brazil, the United States and Canada have demonstrated thatcontrolled-use policy is effective in eliminating the health risks attributable to the modern uses ofchrysotile. Controlled use policy is less restrictive than a ban. Trade and sales are permitted as longas appropriate safety measures are employed in the manufacture, installation and use of chrysotile-containing products. While complying with safety regulations could be expensive for firms, thedecision of whether or not to use chrysotile or substitutes under the safety regulations should bedetermined by the marketplace and not by government. Given the availability of controlled use policyand its effectiveness in addressing the legitimate public health objective which France wishes toachieve, the ban is inconsistent with Article 2.2 in that it is more trade-restrictive than necessary tofulfil its objective.

(iii) Article 2.4 of the TBT Agreement

4.35 Brazil contends that the French ban is inconsistent with Article 2.4 of the TBT Agreementbecause it ignores appropriate and effective international standards. Article 2.4 obliges France to baseits technical regulations on existing international standards, or on any "parts of them", that would beeffective and appropriate in any given circumstance. France violated this obligation when it bannedchrysotile and chrysotile products, ignoring existing international standards that would have been bothappropriate and effective. To establish that France has not violated Article 2.4, the EC must showthat: (i) there are no international standards that apply to asbestos; (ii) if international standards exist,that the ban is consistent with them; or (iii) if international standards exist and the ban is inconsistentwith them, that the international standards would not have been an effective or appropriate means ofaccomplishing France's stated objective. The EC cannot make such arguments.

4.36 Brazil argues that a number of international standards apply to chrysotile and chrysotileproducts, including ILO Convention 162 and Recommendation 172, on the types of asbestos that canbe used (only chrysotile) and how, and the International Organization for Standardization's (ISO) 7337standard, entitled Chrysotile Cement Products Guidelines for On-Site Work, regarding the properinstallation and use of chrysotile-cement products. The fact that the ISO 7337 standard is anapplicable international standard is beyond doubt. Annexes 1 and 3 to the TBT Agreement expresslyrecognize the authority and status of ISO as an international standard-setting body, and the ISO 7337standard directly governs the primary chrysotile product group. Each of these documents state thatchrysotile products may be manufactured and used, but only under controlled conditions and inmodern applications. Each of the standards sets out specific controls to guarantee the safety ofworkers and end-users. They have been incorporated into Brazilian legislation as well as that of manyother countries, including the United States and Canada. The ban is inconsistent with theseinternational standards because it bans all imports, manufacture, use, etc., of chrysotile and chrysotileproducts, whereas these permit their use in modern applications. They only subject them to safetycontrols. Current international standards provide an effective and appropriate means of fulfillingFrance's stated objective and the EC cannot argue otherwise. ILO and ISO standards are "appropriate"for France's stated objective since they were specifically drafted to protect the health of industrialworkers, the general public and others who may come into contact with asbestos. ILO and ISOstandards would also be "effective" in achieving France's stated objective since they have successfully

WT/DS135/RPage 209

protected human health in economies as diverse as those of Brazil, the United States and Canada. TheEC would be hard pressed to provide evidence of a deterioration in the health of citizens from Brazil,the United States or Canada due to adherence to ILO or ISO standards.

4.37 Brazil states that a closer examination of the term "ineffective and inappropriate means" isjustified. The text of Article 2.4 clarifies that this exception is to be quite narrowly construed andapplied. Were it not to be so, Article 2.4 would be rendered useless.56 Members would all too easilyclaim that the applicable international standard was "inappropriate". Second, the Article providesexamples of the situations in which exceptions to the use of international standards are allowed.These include when an international standard would be ineffective or inappropriate due tofundamental climatic or geographical factors or fundamental technological problems. Thus, aMember may ignore an international standard only if the standard will not achieve the results it seeksbecause of its unique conditions in terms of climate, geography, or its economy (i.e. level oftechnological development). No such conditions exist in France. The EC would be unable to presentany evidence to suggest that different conditions apply to France so as to make the standards followedby Brazil, the United States and Canada inappropriate or ineffective for it. France ignored ILO andISO standards because it wished to ban chrysotile to appease public opinion and advantagedomestically-produced and substitute products.

(iv) Article 2.8 of the TBT Agreement

4.38 Brazil argues that the French ban is inconsistent with Article 2.8 of the TBT Agreementbecause it establishes design requirements for products. France's ban is inconsistent with thisobligation because, by prohibiting chrysotile and its use in any product, the ban sets out animpermissible "design or descriptive characteristic". To establish that France has not violatedArticle 2.8, the EC must demonstrate that (i) the ban is a performance requirement; or, in this case,(ii) that adopting a performance requirement would not have been "appropriate". The Communitiescan demonstrate neither of these points. The ban sets out an impermissible "design or descriptivecharacteristic" because it regulates on the basis of the content and description of a product. France hasbanned chrysotile and products containing it but has not banned competing fibres and products thatcontain them. Therefore, the ban advantages French-produced substitute fibres, products which are"like"57 chrysotile, and chrysotile containing products. The ban does not contain regulations based onthe performance of a product. Rather, it states that certain products may be imported and sold only ifthey do not contain a certain input, namely chrysotile. Article 2.8 obliges France to adopt aperformance requirement "whenever possible". In the case of chrysotile, France could have adoptedany of a number of performance requirements that would have enabled it to achieve its statedobjective.

4.39 According to Brazil, France could have adopted, for example, detailed regulations regardingthe importation, production, modern use and disposal of chrysotile and substitute fibres and theirproducts (as France had previously done and as do Brazil, the United States, Canada and many othercountries). Alternatively, France could have established a single, never-to-be-exceeded exposure levelto apply to the manufacture, use and disposal of chrysotile and substitute fibres, and to their products.France could, and should, have adopted a performance requirement for chrysotile and the productswhich contain it. Instead, it adopted a design or descriptive requirement and violated Article 2.8 ofthe TBT Agreement. Were any other findings to be reached by the Panel, it would allow Membercountries to take the much easier route of banning, rather than regulating, products which they claim

56Brazil notes that interpretations that render a treaty provision null or void, or consign it to "inutility"

are to be avoided whenever possible. See United States – Standards for Reformulated and ConventionalGasoline (20 May 1996), WT/DS2/AB/R, p. 23.

57Brazil notes that paragraphs 4.42-4.43 below demonstrate that the man-made substitute fibres andproducts are like products to chrysotile and chrysotile products.

WT/DS135/RPage 210

create health risk. The TBT Agreement is based on the assumption that certain products present risksand that those risks are to be managed through standards. To allow a Member to ban, instead of toregulate, products due to perceived risks would render the Agreement meaningless.

(b) The General Agreement on Tariffs and Trade

(i) Article XI of the GATT

4.40 Brazil submits that the ban is also inconsistent with GATT Article XI because it is aquantitative restriction that is not permitted by the WTO. The ban includes (i) a prohibition of the salein France of chrysotile and chrysotile products, which is a violation of GATT Article III:4, and (ii) aprohibition of the importation of chrysotile and chrysotile products. In fact, paragraphs I and II ofArticle 1 of the ban prohibit "the import […] of all kinds of asbestos fibres […] whether or not thesesubstances are incorporated into materials, products or devices". The latter aspect violates Article XI.In reference to paragraph 1 of Article XI, Brazil argues that the ban on importation is not a "duty, taxor other charge," but is a "prohibition or restriction" that France has instituted and maintained on theimportation of chrysotile from Brazil. Indeed, the ban is the most restrictive of all quantitativerestrictions in that it sets a quota at the level of zero imports.58 Therefore, the portion of the Decreethat bans imports is inconsistent with Article XI:1.59 Brazil further argues that none of the threeexceptions contained in paragraph 2 of Article XI apply to the ban. Simply put, the ban is an outrightprohibition of all imports, supposedly imposed to protect public health. It is not a "standard orregulation for the classification, grading or marketing" of chrysotile or chrysotile products.60

Moreover, whether chrysotile is a "commodity" within the meaning of this exception is questionable.All three exceptions only relate to agricultural products.

(ii) Article III of the GATT and Article 2.1 of the TBT Agreement

4.41 Brazil argues that the ban is inconsistent with France's national treatment obligations underGATT Article III:4 and Article 2.1 of the TBT Agreement. The national treatment obligations ofArticle III:4 and TBT Article 2.1 are violated when a law, regulation or requirement (or a technicalregulation) that affects the internal sale, offering for sale, purchase, transportation, distribution or useof any imported product, accords less favourable treatment to the imported product than that accordedto "like" domestic products. Each of these criteria are satisfied with respect to the ban. The ban isindisputably a law and its three implementing "Arrêtés" are regulations. For the purpose of Article 2.1of the TBT Agreement, the ban is a technical regulation. Article 1 of the Decree bans, among otherthings, the manufacture, processing, sale, offer for sale, distribution and use of all varieties of asbestosfibres and all asbestos-containing products (except for the few temporary exceptions permitted by itsArticle 2). Thus, it indisputably meets the second criterion for the application of GATT Article III:4

58The fact that Article XI applies to the ban is further confirmed by Article XI:2(b) which applies to

"import […] prohibitions," among other restrictions. See also Japan – Trade in Semi-Conductors, L/6309,adopted 4 May 1988, BISD 35S/126, para. 104 (finding Article XI:1 "comprehensive" and applicable to all typesof non-tariff prohibitions).

59The applicability of Article XI:1 to such circumstances has been confirmed by various panels underthe GATT 1947 and GATT 1994. See, e.g., United States Manufacturing Clause, L/5609, adopted 15/16 May1984, BISD 31S/74, 88, para. 34; Japan – Trade in Semi-Conductors, L/6309, adopted 4 May 1988,BISD 35S/116, 152-53, para. 102; United States – Import Prohibition of Certain Shrimp and Shrimp Products,WT/DS58/R (15 May , 1998), paras. 7.11 to 7.17.

60Brazil notes that a GATT Panel has held that a ban (which, of course, precludes marketing) is not"related" to marketing under Article XI:2(b). See Canada - Measures Affecting Exports of Unprocessed Herringand Salmon, L/6268, adopted 22 March 1988, BISD 35S/98, 112, paras. 4.2-4.3 (rejecting Canadian argumentthat a ban on exports of certain unprocessed fish was related to marketing, and finding that, to fall underException Two, the regulation in question must apply to "marketing as such," and that Exception Two does notapply to just any regulation facilitating foreign sales).

WT/DS135/RPage 211

and TBT Article 2.1. It provides less-favourable treatment to chrysotile and chrysotile-containingproducts (which, prior to the ban, were imported from Brazil), than that which it does to Frenchproducts used as asbestos substitutes (and which are not banned).61

4.42 Finally, the ban itself recognizes that the so-called "substitute fibres", and productsincorporating them, are "like" chrysotile and chrysotile-containing products. The few exemptionspermitted by Article 2 of the ban apply when no substitute fibre is equivalent in terms of its end-use tochrysotile.62 In other words, whenever a French substitute fibre can replace chrysotile, chrysotile isbanned. There can be no more convincing proof that chrysotile and substitute fibres are "like"products. Even if the ban did not by its own terms prove the "likeness" of French substitute fibres toimported chrysotile, analysis of the precedents set under GATT demonstrate that chrysotile andsubstitute fibres are indeed alike, as are chrysotile and substitute fibre-containing products. Using thecriteria identified by the Appellate Body in the case on Taxes on Alcoholic Beverages for establishing"likeness"63, it is self-evident that the end uses of chrysotile and substitute fibres are the same. Thefibres are used solely because they emulate the desired characteristics of chrysotile in particularproducts. With regard to "consumers' tastes and habits", chrysotile and substitute fibres are notconsumer goods. They are used solely as inputs in certain products (primarily in various cementproducts today). Industrial consumers purchase substitute fibres rather than chrysotile based onconsiderations of cost and availability. They can do so because substitute fibres are intended toemulate chrysotile's characteristics.

4.43 Brazil asserts that the same reasoning applies to the assessment of the products' properties,nature and quality. Substitute fibres are "like" chrysotile precisely because they emulate chrysotile'scharacteristics. An additional criterion to determine likeness was added after Border Tax decision –tariff classification.64 As previously noted, almost all chrysotile is used as an input into variouscement products. Chrysotile and other fibre-cement products are classified under the sameHarmonized Tariff System heading (which is number 68.11). In all instances, the six and eight digitclassification of chrysotile and other fibre-containing cement products are the same. Therefore,France's conduct violates GATT Article III:4 and the TBT Agreement's Article 2.1, and is inconsistentwith France's national treatment obligation.

(iii) Article I of the GATT and Article 2.1 of the TBT Agreement

4.44 The most-favoured-nation obligations of Articles I:165 and 2.1 are violated "with respect to allmatters referred to in paragraphs 2 and 4 of Article III" (or, for purposes of the TBT Agreement,Article 2.1), whenever any "advantage, favour, privilege or immunity" is granted to a product fromone country and is not "accorded immediately and unconditionally" to a "like product" from otherWTO Members. This happens to be the case with the French ban. As previously demonstrated,Brazil argues that the ban violates GATT Article III:4 and, for the purposes of Article 2.1 of theTBT Agreement, is a technical regulation. The fact that substitute fibres may be imported into France

61According to Brazil, the absence of imports because of the imposition of a ban does not provide a

valid basis for asserting that GATT Article III:4 (and TBT Article 2.1) cannot be applied. Interpretations thatrender a treaty provision null or void, or consign it to "inutility" are to be avoided whenever possible. SeeUnited States – Standards for Reformulated and Conventional Gasoline (20 May 1996), WT/DS2/AB/R, p. 23.

62Brazil notes that the EC acknowledges this when explaining that the French ban does not includechrysotile diaphragms for use in chlorine environments because substitutes cannot safely be used.

63Japan - Taxes on Alcoholic Beverages (4 October 1996), WT/DS8/AB/R, p. 20, quoting Report of theWorking Party on Border Tax Adjustments (2 December 1970) BISD 18S/87, 102, para. 18.

64Brazil notes that this criterion was first cited in EEC – Measures on Animal Feed Proteins, L/4599,adopted 14 March 1978, BISD 25S/49, 63, para. 4.2.

65Brazil recognizes that Canada has not alleged a violation of GATT Article I:1. However, asdemonstrated, the French ban violates the most-favoured-nation obligations of both that Article and ofArticle 2.1 of the TBT Agreement.

WT/DS135/RPage 212

while chrysotile imports are banned constitutes an "advantage, favour, privilege or immunity." Thisadvantage is accorded to imported substitute fibres but is denied to imported chrysotile, which isbanned.

(iv) Article XX of the GATT

4.45 Brazil argues that the General Exceptions of Article XX do not excuse the Decree. To obtainan exception under Article XX, the EC must establish that (i) the ban does not "constitute a means ofarbitrary or unjustifiable discrimination between countries where the same conditions prevail",(ii) that it is not a "disguised restriction on international trade", and (iii) that it is "necessary to protecthuman life or health." The EC cannot argue that the ban meets these conditions. As demonstratedabove, the ban discriminates between like products, without advancing its stated objective. Therefore,it is a "means of arbitrary or unjustifiable discrimination". Also, it disadvantages imports ofchrysotile, but not imports of man-made fibres. Countries like Brazil (and Canada) produce bothchrysotile and substitute fibres. Therefore, the criterion of "discrimination between countries wherethe same conditions prevail" is obviously satisfied. Similarly, as previously demonstrated, the ban is a"disguised restriction on international trade". Although it masquerades as a measure designed toprotect public health, it is an outright product ban that is designed to quell public outrage andadvantage domestic and European manufacturers of substitute fibres and products. Moreover, itcannot be argued that the ban is "necessary" to protect human life or health. For these reasons, the ECshould not be granted recourse to Article XX.

B. UNITED STATES

1. Introduction

4.46 The United States' submission first discusses the facts concerning the health risks of exposureto chrysotile asbestos, and reduction of these risks through regulation. In this connection, theUnited States supplies information correcting certain errors and mischaracterizations in Canada’sdescription of the United States regulation of asbestos and the former United States ban andphaseout on asbestos-containing products. United States regulations are not at issue in thisproceeding. The United States' submission nevertheless seeks to set the record straight because ofCanada’s assertions concerning United States policy. Following a factual discussion, it addresses thelegal provisions that the Panel has been asked to interpret.

4.47 In the view of the United States, chrysotile asbestos is a toxic material that presents a seriousrisk to human health. Chrysotile asbestos is no less toxic than other forms of asbestos. A regulatoryapproach that treats all forms of asbestos on a par with each other is scientifically justified. France,like all other Members of the WTO, has the right to set its own desired level of protection against risksarising from exposure to asbestos, and its regulation on asbestos appears neither discriminatory norunnecessarily trade restrictive in ensuring that level of protection. The United States currently relieson specified work practices and other controls (including a limited ban) to reduce the risk to humanhealth from asbestos exposure. However, the United States does not consider its approach to be theonly appropriate one for regulation of asbestos. Specification of work practices and other controlsdoes not avoid all the risks associated with a hazardous material such as chrysotile asbestos. First,"controlled use" does not eliminate all the risks associated with asbestos. Although it is generally truethat asbestos contained in a cement matrix does not present substantial risks while that product isintact, the same is not true during the production, installation, maintenance, removal, or disposal ofthat product. Second, in many cases a matrix containing asbestos does not remain intact during itsuseful life. Moreover, while the bulk of Canada’s submission focuses on cement-matrix applications,it also acknowledges that chrysotile asbestos is currently used in brake linings and spun fibres for theproduction of insulating tissues or cords. Significant health risks attend the manufacture and repair ofsuch substances. Finally, even the best work practice is effective only to the extent that it is followed;

WT/DS135/RPage 213

accidents, use of improper techniques, and intentional non-compliance are virtually inevitable in theuse of these products. For these reasons, France's ban on the manufacture, processing, distribution incommerce, export, import and sale of asbestos and its products appears to be a WTO-consistentresponse to the risks posed by the use of asbestos.

4.48 As for the legal issues: In the view of the United States, Canada has not met its burden ofproof with respect to any violation by the French Decree of provisions of the GATT or the Agreementon Technical Barriers to Trade ("TBT Agreement"). In particular, Canada has not shown thatimported asbestos and asbestos-containing products are "like products" with respect to substitutefibres and products containing them which are of French origin. As a finding that these products arenot "like products" eliminates any violation of Article III:4, and Article XI:1 is simply irrelevant to theanalysis of this measure, there does not appear to be any violation of the GATT 1994. With respectto the TBT Agreement, the United States disagrees with the EC position that the TBT Agreement isinapplicable to the French Decree. The Panel should reject the EC’s position and find that the Decreeis a "technical regulation" within the meaning of Annex 1 of the Agreement; any other result willopen up a loophole which could entirely nullify the TBT Agreement. Nevertheless, in the UnitedStates view Canada has not proven any violation of Articles 2.2, 2.4, 2.8 or 2.1 of the Agreement.Finally, Canada has not met the particularly high burden of proof for cases of non-violationnullification and impairment.

2. Factual Aspects

4.49 The United States argues that asbestos - whether chrysotile or in other forms66 - is a toxicsubstance. In United States lexicography, it is a "Class A carcinogen", meaning a substance whosecarcinogenic properties have been proven conclusively.67 The IPCS has reached the same conclusion:"[E]xposure to chrysotile asbestos poses increased risks for asbestosis, lung cancer and mesotheliomain a dose-dependent manner."68 The IPCS Report also concludes: "[C]ommercial grades of chrysotilehave been associated with an increased risk of pneumoconiosis, lung cancer and mesothelioma innumerous epidemiological studies of exposed workers."69 In regulating asbestos, the United Statestreats chrysotile asbestos the same as any other recognized form of the substance.70 The findingspresented by Stayner et al.71 support the decision not to distinguish between chrysotile and other formsof asbestos. This study concluded that it is prudent to treat chrysotile with virtually the same level ofconcern as the amphibole forms of asbestos, based on the evidence of a significant lung cancer risk,the fact that workers are generally exposed to a mixture of fibres, and the lack of conclusive evidencefor the "amphibole hypothesis".72 More recent confirmation of the hazardous nature of chrysotile was

66The United States notes that its arguments focus on chrysotile asbestos, as that is the subject of the

Canadian challenge.67United States Environmental Protection Agency ("EPA"), Integrated Risk Information System (IRIS),

Asbestos Substance File (1993) (www.epa.gov/ngispgm3/iris/subst/0371.htm#II) (includes summary of weight-of evidence classification and human carcinogenicity data, including data showing the carcinogenicity ofchrysotile asbestos).

68IPCS Environmental Health Criteria 203 – Chrysotile Asbestos, WHO, 1998, p. 144. (The IPCSdocument cites numerous studies supporting this conclusion).

69IPCS Environmental Health Criteria 203 – Chrysotile Asbestos, WHO, 1998, p. 7.70Airborne Asbestos Health Assessment Update, p. 118 (EPA, June 1986) (concluding that "while

differences in pleural mesothelioma risk attributable to fibre type may exist, they are much less than differencesattributable to other factors").

71Stayner, L. T., Dankovic, D. A., and Lemen, R. A., Occupational Exposure to Chrysotile Asbestosand Cancer Risk: A Review of the Amphibole Hypothesis, 86 American Journal of Public Health, 179-186,1996.

72The United States notes that the "amphibole hypothesis" postulates that the mesotheliomas among theworkers exposed to chrysotile may be explained by confounding exposures to amphiboles, and that chrysotilemay have a lower carcinogenic potency than amphiboles.

WT/DS135/RPage 214

provided by Landrigan, concluding on the basis of an epidemiological study undertaken in Quebecthat "chrysotile asbestos is still indisputably a human carcinogen.73" Concerning exposure to asbestos,IARC stated in 1976 that "at present it is not possible to assess whether there is a level of exposure inhumans [to asbestos] below which an increased risk of cancer would not occur."74 The IPCSreaffirmed this conclusion specifically with regard to chrysotile asbestos in 1998, stating: "[N]othreshold has been identified for carcinogenic risks" with regard to chrysotile asbestos.75 That meansthat it cannot be assumed that any exposure, no matter how small, to asbestos is safe. Canadaquestions France's scientific approach, attacking the use of a "linear risk model". The United Statestakes issue with Canada's criticism of the INSERM Report's use of a linear dose-response model toestimate cancer risk. The use of such a model is entirely appropriate when it comes to estimating therisk of cancer from exposure to asbestos.

4.50 The United States notes that it is not in a position to draw definitive conclusions concerningthe regulatory process in France and the actual factual basis for the French Decree. However,generally speaking, regulatory decision-making relating to carcinogens involves two components:risk assessment and risk management. Risk assessment defines the adverse health consequences ofexposure to toxic agents. Risk management combines the risk assessment with the directives ofregulatory legislation, together with socioeconomic, technical, political, and other considerations, toreach a decision as to whether or how much to control future exposure to the suspected toxic agents.76

Risk assessments are carried out independently from considerations of the consequences of regulatoryaction.77 A risk assessment involves, among other things, quantitative and/or qualitative estimation ofrisks associated with low levels of exposure to carcinogens. While it is always preferable to rely onhuman data, epidemiological studies are often either not available or sufficiently definitive,particularly regarding the specific exposure levels involved, and thus often cannot be relied upon asthe sole basis for a risk assessment.78 In addition, because testing of thousands of animals would benecessary in order to have the sensitivity to detect any but large effects, it generally is not practical tomeasure risks at low exposure levels directly in animal experiments.79 Accordingly, a number ofmathematical models have been developed to extrapolate from high dose animal studies to low humandoses.80

4.51 In the United States, models or procedures that incorporate low-dose linearity have beenadopted when data and information are limited and when there is uncertainty regarding the mechanismof carcinogenic action.81 While low-dose linearity may not be appropriate for all carcinogenic risk

73Landrigan, P. L., Asbestos - Still a Carcinogen, 338 New England Journal of Medicine 1619

(28 May 1998).74Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Man, Vol. 14, IARC,

1976, p. 81.75IPCS Environmental Health Criteria 203 – Chrysotile Asbestos, WHO, 1998, p. 144.76Final Guidelines for Carcinogen Risk Assessment, 51 Federal Register 33992, 33993, col. 3

(EPA, 24 Sept. 1986).7751 Federal Register 33992, 33993, col.3 (24 September 1986).78The United States notes that, as the 1986 EPA carcinogen risk assessment guidelines point out: "It

should be recognized that epidemiological studies are inherently capable of detecting only comparatively largeincreases in the relative risk of cancer. Negative results from such studies cannot prove the absence ofcarcinogenic action … ". (51 Federal Register 33992 (24 September 1986), pp. 33995-96). Canada’s statementthat "no epidemiological study to date has detected a higher health risk [than the linear risk model] resultingfrom low-level exposures" must be viewed in this light.

7951 Federal Register 33992 (24 September 1986), p. 33993, col. 3.8051 Federal Register 33992 (24 September 1986), p. 33997. See also EPA Proposed guidelines for

carcinogen risk assessment, 61 Federal Register 17960, 17962 (23 April 1996). Although these most recentguidelines are not yet final, they demonstrate that EPA’s reassessment of the issues is similar to the approachtaken previously.

8151 Federal Register 33992 (24 September 1986), p. 33997, col.3.

WT/DS135/RPage 215

assessment, it is commonly used in the United States as a default methodology. This methodology issupported by scientific studies and is a reasonably protective approach in the face of uncertainty.82

The use of a linear model is appropriate for a quantitative estimation of the risks associated with lowlevels of exposure to asbestos because of the observed linearity of the response in occupationalstudies. The United States has adopted this approach, in addition, because of the incompleteunderstanding of how asbestos causes diseases in humans.83 In assessing the risk from asbestos, EPAnotes that "[d]irect evidence for linearity of response with asbestos exposure is available from sevenstudies (two of the same plant) that compared lung cancer mortality to the cumulative total dustexposure in asbestos workplaces"[citations omitted].84 Similarly, the limited data that exist formesothelioma also indicate a linear relationship.85 The IPCS states: "there was a clear dose-responserelationship, with crude rates of mesotheliomas (cases/1000 person-years) ranging from 0.15 for thosewith cumulative exposures less than 3530 million particles per cubic meter years … to 0.97 for thosewith exposures of more than 10,590 mpcm-years […].86" After identifying and defining the adverseeffects of asbestos through the risk assessment process, the next step is to make risk managementdecisions. A risk management decision, while taking into account the scientific findings of the riskassessment process, also includes a country's choice on whether and how much to regulate a toxicagent. It is at this stage that a country selects measures and regulations that will achieve its chosenlevel of protection relating to the health of its people.

4.52 In its arguments, Canada has referred to United States regulations concerning asbestos.Because its description of the U.S. regulatory approach is substantially inaccurate, the United Statesproceeds to set the record straight. The U.S. regulatory approach at present includes a mix of controlmeasures, including bans and required work practices. This approach involves a number of complexstatutes, some of which require the consideration of cost, feasibility and other factors besides humanhealth. Almost all control measures are designed to protect workers and building occupants fromexposures resulting from contact with asbestos in installed products. Although France's approach tothe same problem is different, this different approach is also reasonable under the circumstances.

4.53 Canada references the 1989 rule promulgated by the EPA prohibiting the future manufacture,importation, processing and distribution in commerce of asbestos in almost all products ("the AsbestosBan and Phase-Out Rule").87 Several of Canada's statements on this point are factually inaccurate.According to the United States, the Asbestos Ban and Phase-Out Rule was in large part vacated andremanded to EPA by the United States Court of Appeals for the Fifth Circuit in a case entitledCorrosion Proof Fittings vs. Environmental Protection Agency88, based on the court's view that EPAhad not appropriately addressed cost-benefit issues. Contrary to the claims made by Canada that EPAwas incapable of scientifically justifying its ban, and that the risks posed by asbestos were notsupported by scientific facts, the court specifically agreed with EPA's scientific judgment byrecognizing that "[a]sbestos is a toxic material, and occupational exposure to asbestos dust can resultin mesothelioma, asbestosis, and lung cancer."89 Indeed, in the record of rulemaking, EPA provided,among other things, a number of scientific studies and reports on the health risks of asbestos,including: Airborne Asbestos Health Assessment Update90, Report to the United States ConsumerProduct Safety Commission by the Chronic Hazard Advisory Panel on Asbestos91, Asbestiform

8261 Federal Register 17960 (23 April 1996), p. 17965.83IPCS Environmental Health Criteria 203 – Chrysotile Asbestos, WHO, 1998, p. 7.84Airborne Asbestos Health Assessment Update, EPA, June 1986, p. 23.85Airborne Asbestos Health Assessment Update, EPA, June 1986, pp. 23-30.86IPCS Environmental Health Criteria 203 – Chrysotile Asbestos, WHO, 1998, p. 8.8754 Federal Register 29460-29513, 12 July 1989.88Corrosion Proof Fittings v. Environmental Protection Agency, 947 F.2d 1201 (5th Cir. 1991).89Ibid., p. 1207.90Airborne Asbestos Health Assessment Update (EPA, June 1986).91Chronic Hazard Advisory Panel on Asbestos, U.S. Consumer Product Safety Commission, July 1983.

WT/DS135/RPage 216

Fibres: Non-occupational Health Risks92 and "Short-term asbestos work exposure and long-termobservation.93" These studies and reports were discussed in the preamble to the Rule.94 The courtbased its decision on procedural flaws in the EPA rulemaking process and on the court's owninterpretation of the applicable United States statutory risk/benefit balancing standard forpromulgating such rules, not on any disagreement with EPA's findings concerning the health hazardsposed by asbestos. After the judicial remand, EPA imposed a more limited ban on asbestos-containing products, including a ban on any new uses of asbestos.95 That ban remains in place.

4.54 The United States notes that Canada makes much of the argument that asbestos entrained in acement matrix does not present "detectable risk.96" However, Canada's focus ignores the riskspresented by asbestos products throughout their life-cycle. The most significant sources of exposureto asbestos, and therefore risk from asbestos-containing products derive from their manufacture,installation, repair, removal, and disposal, including disposal of products containing asbestos in acement or resin matrix. Moreover, while Canada appears to acknowledge France's concern aboutprotecting health in general terms, it appears to disagree about how protective France should be. Byarguing that certain small risks are equivalent to zero risks, the Canadian submission implicitlyquestions the sovereign authority of a WTO Member to determine the appropriate level of protectionfor its citizens. What Canada or the United States might consider to be adequate protection in aparticular context is not necessarily what other countries must choose. To put it another way, Canadaconcedes that a ban is acceptable for "certain uses where exposure cannot be controlled to anacceptable degree". The United States agrees, but submits that it is up to each Member to determinewhat that "acceptable degree" is. Canada indicates that among the most important commercialapplications of asbestos are as part of brake linings or clutches and in the form of spun fibres for theproduction of insulating tissues or cords. In connection with its analysis of "friction products", whichinclude brake linings and clutches, the United States court in the Corrosion Proof Fittings caserecognized that "[w]orkers are exposed to asbestos during the manufacture, use, repair and disposal ofthese products" and that, in the asbestos ban and phase-out rule, "EPA demonstrates that thepopulation exposure to asbestos in this area is great."97 The court agreed with EPA's determinationthat friction products containing asbestos pose a risk to human health.98

4.55 With respect to Canada's argument that a major commercial application for asbestos is as areinforcement material for cement, plastic, or rubber, the United States asserts that, in theAsbestos Ban and Phase-Out Rule, EPA made certain determinations respecting worker exposure tothese products that were not questioned by the Corrosion Proof Fittings court. EPA determined thatthe manufacture, installation, repair, and disposal of flat and corrugated asbestos-cement sheet expose

92Asbestiform Fibres: Non-Occupational Health Risks, NAS, NRC, 1984.93Seidman, H., Selikoff, I .J., Hammond E. C., Short-Term Asbestos Work Exposure and Long-Term

Observation, 330 Annals of the New York Academy of Sciences 61-89, 1979.9454 Federal Register 29460 (12 July 1989), p. 29468-70.9540 Code of Federal Regulations (CFR) 763.165-763.169 (59 FR 33208, 28 June 1994).96The United States notes that it is not entirely clear what Canada means by the term "undetectable

risk". The presence of asbestos fibres in the air or other media can be detectable or undetectable. A risk can besignificant, insignificant, or non-existent. It appears that Canada uses the term "undetectable risk" to refer to arisk that Canada deems insignificant. Significance, however, is a judgment call that can only be made by theregulatory authority responsible for public health and safety. It is up to France to determine what level of risk tothe French people from asbestos (or any other hazard) is significant.

97Corrosion Proof Fittings v. Environmental Protection Agency, 947 F.2d 1201 (5th Cir. 1991), p. 1224.98Significant damages have been awarded in U.S. courts with respect to brake applications of asbestos.

In 1985, a retired brake mechanic who was dying of mesothelioma won a verdict of $2 million in a court actionagainst Raybestos Manhattan. See McDonald AD, et al., Dust Exposure and Mortality in an AmericanChrysotile Asbestos Friction Products Plant, 41 Br J Ind Med 151-157, 1984; Newhouse M. L. andSullivan K. R., A Mortality Study of Workers Manufacturing Friction Materials: 1941-86, 46 Br J Ind. Med,176-179, 1989.

WT/DS135/RPage 217

workers to asbestos.99 Similarly, EPA determined that the manufacture and installation of asbestos-cement pipe provide "primary routes of exposure" of workers to asbestos from these products, andworkers may also be exposed during the removal of asbestos-cement pipe.100 The United Statesgenerally agrees with Canada's assertion that as long as asbestos is held within a cement or resinmatrix in an undisturbed state, there is minimal exposure to the fibres - but only while the matrixretains its integrity. Much asbestos has been installed in United States buildings. Because of the highhealth risk from disturbed building materials and the reduced risk from intact asbestos-containingmaterials, EPA has issued guidance recommending management of asbestos-containing materials inplace.101 Unfortunately, cement and resin matrices do not remain undisturbed. Putting asidesignificant releases that occur during the manufacturing process102, releases of asbestos can occurwhen, for example, asbestos-cement pipes are installed (which requires cutting the material), andwhen asbestos-containing material (such as cement) deteriorates, as through peeling, cracking, orcrumbling. Fibre release could also result when the material is dry, has the capacity to be crumbled,pulverized or reduced to powder by hand pressure, or is subject to sanding, grinding, cutting orabrading.103

4.56 The United States asserts that a cement matrix in which asbestos is bound can undergo anatural process of erosion or degradation resulting in asbestos fibre release: "[T]he release of fibresfrom external asbestos-cement products [such as siding] due to weathering can be an importantexternal source of asbestos contamination that can be carried into or can infiltrate into the buildingenvironment". This has been acknowledged by the 1991 Health Effects Institute-Asbestos Research(HEI-AR) report, Asbestos in Public and Commercial Buildings: A Literature Review and Synthesisof Current Knowledge.104 As reported in the HEI-AR document, researchers found that weatheredasbestos-cement sheet products washed from gutters and onto walkways were an important source ofchrysotile carried by foot or wind into a classroom.105 The report also cited a research finding ofincreased ambient air concentrations in the vicinity of buildings with asbestos-cement products ontheir exterior.106 Mere maintenance of some asbestos bound in a matrix may disturb the matrix andthereby create additional exposures to the asbestos fibres. For example, in an EPA study107 conductedat 17 schools in New Jersey involving spray buffing of resilient floor tile containing asbestos, airborneasbestos concentrations were approximately five times higher during than before spray-buffing withhigh-speed machines, whereas spray-buffing with low speed machines showed a two-fold increase.For school maintenance workers, the maximum estimated eight-hour time-weighted average exposureconcentration was 0.093 f/cc. Similarly, routine spray-buffing and wet-stripping as well as ultra-highspeed burnishing and wet-stripping of asbestos-containing resilient floor tile can result in elevated

9954 Federal Register 29460-29513 (12 July 1989), p. 29491.10054 Federal Register 29460-29513 (12 July 1989), pp. 29496-97.101Managing Asbestos in Place: A Building Owner’s Guide to Operations and Maintenance Programs

for Asbestos-Containing Materials, EPA, July 1990.102The United States notes that, for example, a study of mortality among long-term employees of an

Ontario asbestos-cement factory found a substantially increased risk of death from lung cancer andmesothelioma. Finkelstein, M. M., Mortality Among Long-Term Employees of an Ontario [Canada] Asbestos-Cement Factory, 40 Br. J. Ind. Med. 138-44, 1983.

103The United States notes that this has been recognized by EPA’s asbestos National Emission Standardfor Hazardous Air Pollutants (NESHAP), promulgated under §112 of the Clean Air Act, 42 U.S.C. 7412. 55Federal Register 48406, 48408-09 (Nov. 20, 1990), codified at 40 CFR part 61, subpart M.

104Asbestos in Public and Commercial Buildings: A Literature Review and Synthesis of CurrentKnowledge, Health Effects Institute-Asbestos Research Report, 1991, p. 4-32.

105Ibid., pp. 4-32 and 4-33.106Ibid., pp. 4-33.107Project Summary: Airborne Asbestos Concentrations During Buffing of Resilient Floor Tile, EPA,

October 1993, p. 4.

WT/DS135/RPage 218

levels of airborne asbestos.108 Part 2 of the Kominsky study demonstrates that the ultra high-speedburnishing and wet-stripping procedures were associated with a maximum estimated eight-hour time-weighted average exposure concentration of 0.275 f/cc to operations and maintenance staff.109

Likewise, the HEI-AR Report states that: "buffing, wax stripping, and other abrasive treatments maycause the release of particulate material from the surface of the floor tile".110

4.57 The United States' regulatory programme on asbestos is largely aimed at controlling exposureto asbestos when it is no longer bound in a matrix. United States regulations address renovation anddemolition of buildings111 and identification and management of asbestos-containing material inschools.112 Regulations of the Occupational Safety and Health Administration (OSHA) of theDepartment of Labor address worker exposures to asbestos including manufacture, installation,renovation, removal and custodial work where workers come into contact with asbestos-containingmaterial. OSHA has established a permissible exposure limit and mandated extensive work practicecontrols, enclosures, hazard communication, training, medical and other industrial hygiene practices toprotect both workers contacting asbestos and other workers who are nearby. Enforcing and complyingwith these regulations entail a significant commitment of resources by both the public and privatesectors. Even within a cement or resin matrix, the risk from asbestos is not negligible. According toan analysis conducted in 1991 by the Health Effects Institute-Asbestos Research113, janitors,custodians and maintenance workers exposed to ambient asbestos fibre levels of 0.1 f/ml (thepermissible exposure limit currently allowed by OSHA regulations) were subject to an estimatedincreased risk of death from cancer of 2 in 1,000. The same analysis estimated that buildingoccupants (school children and office workers) exposed to airborne asbestos fibres from asbestos-containing materials (presumably many of which, as building materials, would have been encased incement or resin) have a lifetime cancer risk from such asbestos exposure of 4 to 60 per 1,000,000.Contrary to Canada's suggestion, such risks are not equivalent to zero. Each country must determinefor itself what level of protection from risks of exposure to asbestos it wishes to achieve, i.e. whatrisks to its population it is willing to accept. It is not for another country to tell France that certainrisks to its population are not significant. By way of illustration, the United States regulates risks onthe order of 1 in 100,000 (1 x 10-5) or 1 in a million (1 x 10-6) in a number of instances. Canadaconcedes that "[t]he principle of controlled use also means that certain uses for which exposure cannotbe controlled to an acceptable degree would be banned". The discussion above demonstrates thatexposure to asbestos even in a cement or resin matrix cannot be controlled sufficiently to eliminate allrisk.

4.58 The United States argues that both of the parties have mischaracterized the substitutes forasbestos and asbestos-containing products. Some products that currently contain asbestos may bemanufactured simply by removing the asbestos, thus eliminating any substitution of risk.Alternatively, a wide variety of fibrous substances are being used commercially as replacementmaterials for asbestos-containing products. These include the man-made mineral fibres (consisting ofglass fibres, rock wool, slag wool, refractory ceramic fibres), selected organic fibres (e.g., aramid,carbon/graphite, polyolefin), and several naturally occurring mineral fibres other than asbestos(e.g., wollastonite, sepiolite, palygorskite). The potential health effects for these non-asbestos fibres

108Kominsky J. R., Freyberg R. W., Clark P. J., Edwards A; Wilmoth, R. C., Brackett, K. A., AsbestosExposures During Routine Floor Tile Maintenance. Part 1: Spray-Buffing and Wet-Stripping; Part 2: UltraHigh Speed Burnishing and Wet-Stripping, 13 Appl. Occup. Environ Hyg. 101-112 (February 1998).

109Ibid., pp. 107-112110Asbestos in Public and Commercial Buildings: A Literature Review and Synthesis of Current

Knowledge, Health Effects Institute-Asbestos Research Report, 1991, pp. 4-70.111Asbestos NESHAP, 40 CFR 61.145.112Regulations issued under the Asbestos Hazard Emergency Response Act (AHERA), 15 USC 2641 et

seq.: 40 CFR part 763, subpart E.113Asbestos in Public and Commercial Buildings: A Literature Review and Synthesis of Current

Knowledge, Health Effects Institute-Asbestos Research Report, 1991, pp. 8-9 - 8-10.

WT/DS135/RPage 219

have been evaluated by EPA,114 the IARC115 and the IPCS.116 Although limited health effectsinformation exists for many of these fibres, available data do not indicate that these fibres are as toxicas chrysotile asbestos. For example, none of these fibres have been found to cause either malignant ornon-malignant respiratory diseases similar to those associated with asbestos exposure in humans.Unlike asbestos fibres, these substitute fibres have not been classified as carcinogenic to humans orknown human carcinogens. The only fibre that has been shown to be more hazardous than asbestosfibres is erionite. Erionite, however, is not known to be available in commerce at this time.117

4.59 The United States notes that Canada continually urges that "controlled use" will bring the riskassociated with chrysotile asbestos to "undetectable" levels. It also compares the risk from asbestos tothat of other products and activities, concluding that many are more risky than asbestos. Canadaoverstates the efficacy of "controlled use". For example, Canada states that where it is necessary tocut chrysotile-cement materials on site, "the use of tools that almost entirely eliminate emissions (low-speed saws, with water injection or equipped with suction units), and the wearing of a mask by theoperator guarantee their safety". United States regulations recognize, however, that masks may not besufficient in some situations and require the use of a supplied-air respirator which is obviously morecumbersome and costly.118 When a government makes its choice whether to ban a product or to optfor controlled use, it must necessarily take into account the anticipated effect of the regulations on itspopulation. Beyond the obvious point that "almost" eliminating emissions of asbestos fibres is not thesame as eliminating them, it must be acknowledged that 100 per cent compliance with "controlleduse" of asbestos is not a realistic expectation due to the burdensome nature of certain work practicesconcerned. Even the best work practice is effective only to the extent that it is followed; accidents,use of improper techniques, and international non-compliance are virtually inevitable in the use ofthese products.119

4.60 The United States argues that Brazil has made a number of unjustified and inaccurateassertions concerning the U.S. Environmental Protection Agency's former ban on all forms of asbestosand asbestos-containing products in the United States, and a domestic court decision concerning theEPA ban. Of course, this U.S. domestic court did not rule on the consistency of the EPA ban with theTBT Agreement or the GATT. It only dealt with whether EPA had complied with the risk/benefitbalancing requirements of the U.S. Toxic Substances Control Act (TSCA). The risk/benefit standardin this U.S. statute is irrelevant to the Panel's present task of determining whether France's ban onasbestos is consistent with the WTO Agreement. France has not adopted such a risk/benefit balancingstandard as the basis for a ban. First, Brazil misrepresents the domestic court decision and thesituation following, repeating errors made by Canada. The court in 1991 upheld EPA's determinationthat "[a]sbestos is a toxic material, and occupational exposure to asbestos dust can result inmesothelioma, asbestosis, and lung cancer". The court based its decision not on any disagreementwith EPA's findings concerning the health hazards posed by asbestos, but instead on procedural flawsin the EPA rulemaking process and on the Court's own interpretation of the statutory risk/benefit

114Health Hazard Assessment of Non-Asbestos Fibres, EPA, 1988.115Man-Made Mineral Fibres and Radon: Monographs on the Evaluation of Carcinogenic Risks to

Humans, Volume 43, pp. 39, 148-52, IARC 1988.116Asbestos and other Natural Mineral Fibres, IPCS, 1986, Environmental Health Criteria 53,

International Programme on Chemical Safety, World Health Organization, Geneva; Man-Made Mineral Fibres(IPCS 1988), Environmental Health Criteria 77, International Programme on Chemical Safety, World HealthOrganization, Geneva; Selected Synthetic Organic Fibres, (IPCS 1993), Environmental Health Criteria 151,International Programme on Chemical Safety, World Health Organization, Geneva.

117In addition, see the US answer to question 4 of the EC (contained in Annex II, Section II.A.4).11840 CFR 763.121(h) (EPA regulations covering employees of certain state and local governments

conducting asbestos abatement projects); 29 CFR 1926.1101(g)(2)(v) (OSHA asbestos regulations forconstruction).

119In addition, see the US answer to question 2 of the EC (contained in Annex II, Section II.A.4).

WT/DS135/RPage 220

balancing required by TSCA. Because the court agreed with EPA on the health effects of asbestos,EPA did not, after the court decision, need to carry out a "thorough review of scientific and medicaldata" as a basis to authorize or ban the products Brazil has listed. All EPA did, pursuant to the court'sinstructions, was to determine which product categories were no longer manufactured, imported, orprocessed when the rule was issued. For all such products the ban was maintained. EPA also bannednew uses of asbestos.120

4.61 Second, contrary to what is alleged by Brazil, the United States has not determined thatcontrolled-use policy effectively eliminates the health risk attributable to modern-day uses ofchrysotile. Third, Brazil has downplayed the health risk from asbestos by lifting a quotation out ofcontext from the Health Effects Institute report, in a manner that does not accurately reflect thediscussion of mathematical models found in this section of the HEI report.121 The United Statesaddressed these issues as described above at paragraph 4.40. Finally, concerning Brazil's discussionof the French Decree, the United States notes that Brazil concedes that the objective of the Decree -protection of the health of workers and the public - is a legitimate objective within the meaning ofArticle 2.2 of the TBT Agreement. The United States agrees with Brazil on this point but would alsonote that France has a right to set its "legitimate objective" to establish the protection of the health ofFrench workers and consumers at the level it deems appropriate. However, when alleging that there isno "rational link" between the French Decree and health protection, stating that the Decree will notmake those now sick healthy, and removing it would not make any of those now healthy sick, Brazilconveniently omits the function of the Decree in preventing future exposure and future disease thatwould result from that exposure.

3. Legal Aspects

4.62 For the reasons below, the United States suggests that the Panel should find that Canada hasfailed to meet its burden of proof that the French Decree violates any provision of theWTO Agreement. The Panel should also find that Canada has failed to make a showing that theFrench Decree gives rise to non-violation nullification or impairment of benefits accruing to Canada.

(a) The General Agreement on Tariffs and Trade

(i) Article XI of the GATT

4.63 With respect to Canada's argument that the Decree violates Art. XI:1 because it imposes anabsolute prohibition or restriction on imports, the United States agrees with the EC that Article XI issimply not relevant to these proceedings, and the Decree should be analyzed under Article III instead.The Decree regulates characteristics of asbestos and asbestos-containing products. It applies to allasbestos, and is applied to imported products "at the time or point of importation," in the words of theNote Ad Article III.

(ii) Article III of the GATT

4.64 With respect to Canada's allegation that the Decree violates the national treatment obligationsembodied in GATT Article III:4, the United States argues that, to show a violation of Article III, theremust be discrimination - that is, unlike treatment of like products. Yet the relevant domestic andimported products here are not "like products" for the purposes of Article III:4. As the EC has noted,the classic statement of the factors relevant in determining what constitutes a "like product" is found

120For the details, see 40 Code of Federal Regulations 763.160, 763.165-763.169 (59 Federal Register

33208 (28 June 1994).121Health Effects Institute – Asbestos Research, Asbestos in Public and Commercial Buildings:

A Literature Review and Synthesis of Current Knowledge, Cambridge, 1991, pp. 6-9.

WT/DS135/RPage 221

in the Working Party report on Border Tax Adjustments of 1968, which defined like products in termsof "the product's end-uses in a given market; consumers' tastes and habits, which change from countryto country; the product's properties, nature and quality".122 The United States generally agrees withthe EC's analysis that the properties, nature and quality of asbestos and asbestos-containing productson the one hand, and substitutes on the other, are not "like". The substitutes are by definitionsubstitutable for asbestos and asbestos-containing products for certain uses, but that does not meanthat they are "like products".

4.65 According to the United States, Canada has not made the correct product comparison for thepurpose of determining whether the relevant products are "like products" under Article III:4. Inconsidering a regulation that bans asbestos and requires the use of substitutes, the relevant products tocompare are the following: (i) asbestos must be compared to substitute fibres; and (ii) productscontaining asbestos must be compared to products that do not contain asbestos but which perform thesame function. Where the asbestos elements of a product were inessential, the substitute product mayconsist of the same product minus the asbestos element (e.g., a kitchen hot pad with thick cottonpadding but no asbestos); or the same product redesigned to eliminate the need for asbestos; or asimilar product which uses different fibres (e.g., a hot pad made of glass fibre); or a similar productmade of what the EC describes as "classic materials" (e.g., a trivet made of cast iron, ceramic orplastic). The physical correspondence between the two classes of products is therefore considerablyweaker than Canada has assumed. Canada has failed to show that asbestos and asbestos-containingproducts and the substitutes have the same "properties, nature and quality". The known severeadverse human health effects of asbestos are another reason why asbestos-containing products are not"like" the substitutes for which adverse health effects have not been demonstrated. The substitutefibres differ considerably in physical structure and properties from chrysotile asbestos and thus cannotbe considered "like products." For example, while chrysotile is a naturally occurring mineral which iscrystalline in nature, the man-made mineral fibres (MMMF) are amorphous (non-crystalline) silicateswhich are produced from a liquid melt of different starting materials (e.g., slag, natural rock, glass,clays). Furthermore, unlike chrysotile asbestos, MMMF do not split longitudinally into smaller fibrilsof smaller diameter, but may break transversely into shorter segments.123

(iii) Article XXIII:1(b) of the GATT

4.66 The United States argues that Canada has failed to meet the special burden imposed byArticle 26.1 of the Dispute Settlement Understanding on parties making claims of non-violationnullification or impairment. The United States has been one of the strongest proponents of the non-violation remedy, as an essential safeguard for bargained-for market access rights against frustrationby government actions. But the requirements of the non-violation remedy are not satisfied here. Asthe EC have noted, the text of Article XXIII:1(b) establishes three elements that a complaining partymust demonstrate in order to make a cognisable claim under Article XXIII:1(b): (i) application of ameasure by a WTO Member; (ii) a benefit accruing under the relevant agreement; and (iii)nullification or impairment of the benefit as a result of the application of the measure. Canada, as thecomplaining party, has the burden of presenting detailed evidence in support of all three elements. Inthe present case, there is no dispute that the French Decree is a measure of a Member. The question issimply whether Canada has a legitimate expectation of benefits accruing to it. The precedents areclear that for expectations to be legitimate, they must take into account all measures of the partymaking the concession that could reasonably have been anticipated at the time of the concession.

122BISD 18S/102, para. 18.123Man-Made Mineral Fibres, IPCS Environmental Health Criteria 77, 1988, pp. 11-12; Man-Made

Mineral Fibres and Radon: Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 43, IARC,1988, pp. 39-53; Asbestos and Other Natural Mineral Fibres, Environmental Health Criteria 53, IPCS, 1986,pp. 22-24.

WT/DS135/RPage 222

4.67 The United States considers that, as a matter of principle, the Panel should reject thepossibility of a finding of non-violation nullification or impairment with respect to health and safetyregulations that respond to the development of scientific knowledge concerning health risks.Members do not have a legitimate expectation that regulatory measures will stay static in the face ofexpanding scientific knowledge concerning health risks, and changing societal decisions concerningthe level of acceptable risk. Canada is also in a poor position to argue that an asbestos ban wasunforeseeable at the time it negotiated the tariff concessions on asbestos. The dangers posed to humanhealth by asbestos are notorious and have been so for many years. Pliny, the ancient Roman author,described the "diseases of slaves" as including exposure to the textile processes of preparing andweaving asbestos, and even referred to the use of a transparent bladder skin as a respirator to avoidinhalation of dusts by slaves.124 As of the time of the first GATT negotiating round in 1947,asbestosis had already (in the 1920s) been identified as a distinct condition caused by asbestos.125

By 1935 asbestosis was widely recognized as a mortal threat affecting a large fraction of those whoregularly worked with the material.126 In addition, by the mid-1940s there were indications thatexposure to asbestos in animals and humans was associated with lung tumours.127 Thus Canadashould have reasonably expected subsequent regulatory action (such as a ban) by a GATT contractingparty as a result. As of the Dillon Round of 1960-61, Canada had even more reason to foresee thepossibility of restrictive regulations of asbestos. An international symposium of experts on thepathogenesis of lung cancer in 1953 published its conclusions and recommendations in a journalwhich editorialised: "[I]t seems to be beyond discussion that cancer of the lung is sometimes causedby occupational exposure to asbestos."128 In addition, two major studies had been published in 1955on cancer in the textile industry demonstrating the relationship between asbestosis and lung cancer.129

Since the early 1960s, the hazards posed by asbestos - and particularly chrysotile asbestos - havebecome even more widely known and documented.130

(b) The Agreement on Technical Barriers to Trade

4.68 With respect to Canada's arguments that the French ban on asbestos is inconsistent with anumber of provisions in the Agreement on Technical Barriers to Trade (TBT), the United Statesargues that Canada has misread the relevant TBT articles. The interpretation of the TBT Agreementon which Canada's arguments are based attempts to read into the Agreement obligations that do notexist. The United States urge the Panel to reject that interpretation. The EC, on the other hand, haveargued that the French Decree is not a "technical regulation" because it is a categorical ban onasbestos and products containing asbestos. They have argued that general bans, and specifically thisproduct ban, are not "technical regulations" because they allegedly do not "lay down productcharacteristics or their related processes and production methods" within the meaning of paragraph 1of Annex 1 of the TBT Agreement. The United States disagree with the EC's view on this point. Inthis instance, the Decree lays down "product characteristics […] with which compliance ismandatory". The characteristics in question are that the product may not contain any asbestos if it isto be marketed, offered for sale, imported, exported, etc. in France. Compliance with the exclusion ofasbestos is mandatory except if the French Government has accorded a derogation, in which caseadherence to the terms of the derogation is mandatory. In any event, the French Decree is a technical

124Castleman B. I., Asbestos: Medical and Legal Aspects, 4th ed., 1996, p. 1.125Lilienfeld, D. E., The Silence: The Asbestos Industry and Early Occupational Cancer Research - A

Case Study, 81 American Journal of Public Health 791, 792 (1991).126Castleman B. I., Asbestos: Medical and Legal Aspects, 4th ed., 1996, p. 39.127Lilienfield, D. E., The Silence: The Asbestos Industry and Early Occupational Cancer Research – A

Case-Study, 81 Am. J. Pub. Health 791, 794 (1991); Castleman B. I., Asbestos: Medical and Legal Aspects, 4thed., 1996 pp. 53-65, 135.

128Castleman B. I., Asbestos: Medical and Legal Aspects, 4th ed., 1996 pp. 94, 95.129Ibid. pp. 97, 98.130See IPCS Environmental Health Criteria 203 – Chrysotile Asbestos, WHO, 1998, and sources cited

therein.

WT/DS135/RPage 223

regulation within the meaning of the TBT Agreement and is subject to the substantive rules of theTBT Agreement. The EC's interpretation of Annex 1 would open up a loophole of potentially hugedimensions in the TBT Agreement. Measures having a very significant impact on trade - for instance,regulations limiting the characteristics of spreadable butter or wool - could simply be redefined asproduct bans. Similarly, the EC argument would mean that a regulation on the safety of infant toysthat excluded any parts below a certain size (to prevent choking) would not be a "technical regulation"nor would regulations excluding water from being added to ham. The provisions of theTBT Agreement would then be rendered a nullity. Such a reading of the TBT Agreement isimpermissible as a matter of treaty interpretation, and undesirable as a matter of trade policy. Thisdoes not mean that the French Decree does not satisfy the requirements of the TBT Agreement,however. As discussed below, although the TBT Agreement applies to the French Decree, Canadahas failed to make a case that the French Decree violates any of the provisions it has cited.

(i) Article 2.1 of the TBT Agreement

4.69 The United States argues that, for the reasons discussed in relation to Article III of the GATT,these products are not "like products." Furthermore, since the ban is applied without discrimination asto the source of the product, discrimination between foreign sources is not an issue.

(ii) Article 2.2 of the TBT Agreement

4.70 The United States notes that Article 2.2 provides a key element of the disciplines in theTBT Agreement. From the standpoint of the United States, certain aspects of Article 2.2 areparticularly important with respect to health and safety regulation. The first sentence of Article 2.2 isimportant because it recognizes that in certain circumstances, technical regulations may createnecessary obstacles to trade, and the creation of such necessary obstacles is consistent with theTBT Agreement. We note that the "legitimate objectives" enumerated (non-exhaustively) inArticle 2.2 specifically include protection of human health or safety. Article 2.2 also recognizes thatin assessing the risks that may arise from non-fulfilment of a legitimate objective, a government mayconsider a number of elements, including available scientific and technical information, relatedprocessing technology or the intended end-uses of a product.

4.71 The obligation in Article 2.2 that technical regulations are not to be more trade restrictive thannecessary to fulfill a legitimate objective should be interpreted in a manner similar to Article 5.6 of theAgreement on Sanitary and Phytosanitary Measures (SPS).131 Such a reading is supported by the sixthclause of the Preamble to the TBT Agreement, which provides that: "[R]ecognizing that no countryshould be prevented from taking measures necessary … for the protection of human, animal or plantlife or health, [or] of the environment […] at the levels it considers appropriate […]." TheUnited States argues that the preamble is part of the context of Article 2.2 in the sense of Article 31 ofthe Vienna Convention on the Law of Treaties, and provides an authoritative indication of theTBT Agreement's object and purpose for the purposes of treaty interpretation. Thus, in order for aMember to show that a government's technical regulation is more trade-restrictive than required, itwould need to show that there is another measure that is reasonably available, fulfils the regulatingMember's legitimate objectives, and is significantly less restrictive to trade. Accordingly, thecomplaining party should be required to identify a specific alternative measure that is reasonablyavailable - as a Member is not required to do what is unreasonable. Furthermore, the alternativemeasure must make a significant difference from a trade perspective. There should be no need toadopt an alternative measure if it makes only an insignificant difference in terms of trade. Most

131Article 5.6 provides in relevant part "when establishing or maintaining sanitary or phytosanitary

measures to achieve the appropriate level of sanitary or phytosanitary protection, Members shall ensure that suchmeasures are not more trade-restrictive than required to achieve their appropriate level of sanitary orphytosanitary protection, taking into account technical and economic feasibility".

WT/DS135/RPage 224

importantly, the complaining party must demonstrate that the alternative measure fulfils thegovernment's objectives. Canada has failed to demonstrate that its preferred alternative to the Frenchban – i.e. "controlled use" of asbestos and asbestos products – fulfils the French Government's stated"legitimate objective" of protection of human health.

4.72 Canada alleged that the Decree does not address the "true problem" of asbestos in Francewhich Canada identifies as the flocking of asbestos. Yet it is not for Canada to determine whatFrance's "true problem" is. It is up to France to determine what level of protection to afford itscitizens. Second, Canada alleges that the French Decree violates Article 2.2 because it fails toacknowledge the "scientific reality" that chrysotile encapsulated in a matrix is harmless. Yet asdiscussed above and as demonstrated by the EC, encapsulated asbestos is not harmless at all, as theencapsulation can easily be breached, and is likely to be breached during the product's life cycle,resulting in release of fibres and elevated risk to human health. Canada alleges that the French Decreeviolates Article 2.2 because it replaces use of chrysotile – an allegedly harmless product – withsubstitutes whose health risks are unknown. The United States fundamentally objects to this readingof the TBT Agreement. Canada is implicitly arguing that any regulatory action negatively affectingtrade in a product has to be tested against the hypothetical risks engendered by use of likely alternativeproducts. This test has no basis whatsoever in the TBT Agreement.

(iii) Article 2.4 of the TBT Agreement

4.73 Canada has asserted that Article 2.4 requires a panel to determine: (i) whether a technicalregulation on chrysotile is required; (ii) whether there are international standards concerningchrysotile, (iii) whether the international standards are effective and appropriate to achieve theobjective; and (iv) whether the Decree is based on international standards. Under this analysis,Canada concludes that France adopted the most restrictive measure possible despite the fact that theinternational community has developed standards representing a less restrictive approach(i.e., controlled use). This analysis misreads Article 2.4. First and fundamentally, Article 2.4 does notcontemplate that a panel determine whether a technical regulation is or is not required. The burden ofproof is on Canada to demonstrate that international standards exist and are relevant. In regard to theILO standard, both the ILO Convention 162 and Recommendation 172 allow participating countries tochoose the approach they find to be appropriate to protect workers from asbestos hazards. Indeed, theProvisional Record to the 72nd Session of the International Labour Conference, which adoptedConvention 162, states concerning Article 10 of Convention 162 (which Canada now claims tocondition a ban on a finding concerning the risk posed by product substitutes): "The Governmentmember of Canada saw nothing in Article 10 that would prevent any country from doing whatever itwanted with respect to asbestos." 132

4.74 The United States argues that Brazil's interpretation of Article 2.4 of the TBT Agreementignores the fact that climate, geography and fundamental technological problems are listed asexamples, not an exhaustive list, of reasons that an international standard may be an "ineffective andinappropriate means" of achieving a Member's legitimate objective.

C. ZIMBABWE

1. Introduction

4.75 As an important producer and exporter of chrysotile (white) asbestos fibre and productscontaining chrysotile asbestos and also as a developing country in need of foreign exchange,Zimbabwe argues that it has a substantial interest in the outcome of this proceeding. In fact, the

132International Labour Conference Provisional Record, 72nd Session, Geneva, 1986, 29/1: Fourth

Item on the Agenda: Safety in the Use of Asbestos, pp. 29/8.

WT/DS135/RPage 225

present dispute is of such importance to Zimbabwe's asbestos industry and indeed its whole economythat the Government of Zimbabwe has decided, for the first time ever, to have recourse to the disputesettlement mechanism of the WTO. Zimbabwe is of the view that the ban of chrysotile asbestos andproducts containing chrysotile asbestos by France is unjustified and contrary to relevant rules of theWorld Trade Organization (WTO). The ban should therefore be lifted without delay. Zimbabwebelieves that it is not incumbent upon it as a third party to this dispute to set out in full the case againstthe responding party, i.e. the EC. Accordingly, Zimbabwe will limit itself in this submission to thePanel to addressing a number of factual and legal aspects of this dispute that it feels are of particularimportance to the outcome of this proceeding. Zimbabwe argues that the complaining party in thiscase, i.e. Canada, has made a compelling case with respect to both the factual and legal issues indispute as to why the ban on chrysotile asbestos and products containing chrysotile asbestos isinconsistent with relevant WTO rules and must be withdrawn immediately.

2. Factual Aspects

4.76 Zimbabwe asserts that the chrysotile asbestos industry is of great economic importance to itseconomy. Zimbabwe ranks among the world's largest producers of chrysotile asbestos. In Africa,Zimbabwe is the number one producer of chrysotile asbestos. It produces a high-quality chrysotileasbestos fibre and has sufficient underground reserves for at least another 25 years and infrastructureto continue operations for many more years to come. Chrysotile asbestos currently accounts forabout 18 per cent of Zimbabwe's mineral production index of volume and value. Crocidolite (blue)and amosite (brown) asbestos are not mined in Zimbabwe. As a developing African country,Zimbabwe relies primarily on natural resource products and other primary products for much of itsexport revenue. In terms of the revenue it generates, chrysotile asbestos is second only to gold as faras the mining sector is concerned. As much as 95 per cent of the country's total asbestos fibreproduction is exported. In 1998, for example, 150,000 tonnes of chrysotile asbestos were exported outof a total production of some 175,000 tonnes, generating foreign exchange in excess ofZW$1.5 billion. In addition to the export of chrysotile asbestos fibres, more than 7,500 tonnes ofasbestos-cement products, valued at over ZW$30 million, were exported. Zimbabwe's sole producerof chrysotile asbestos fibre is African Associated Mines. The European Union in general, and Spainand France in particular, have traditionally been important export markets for African AssociatedMines.

4.77 Zimbabwe argues that African Associated Mines suffered a dramatic (more than 50 per cent)drop in its sales to France in 1996. The setback suffered by African Associated Mines in the Frenchmarket is directly attributable to the French Government's actions. It should be pointed out in thisconnection that in mid-1996 the French Government announced its intention to ban asbestos. Beforethat, i.e. towards the end of 1995, the French Government had already announced a programme toreduce the risks associated with exposure to asbestos. There is therefore clear evidence that theFrench ban on asbestos and products containing asbestos has had a direct and damaging impact onZimbabwe's asbestos industry. The significance of the asbestos industry to Zimbabwe cannot beoverstated. The country has immensely benefited from its existence. African Associated Minesdirectly employs about 6,000 people in Zimbabwe, which amounts to about 20 per cent of totalemployment in the mining industry. The industry indirectly sustains more than 70,000 people in andaround the mining towns of Zvishavane and Mashava. There are no other industries in these towns,meaning that a decline of the asbestos industry would cause dislocation with all its attendant socialconsequences. It should be borne in mind in this context that the Zimbabwean economy has facedconsiderable difficulties in the past decade and has not been able to create a sufficient number of newjobs. Out of a labour force of 5 million people, only 1.4 million people are gainfully employed. Apartfrom generating revenue for the Government of Zimbabwe, the asbestos industry has injecteddynamism into the country's economy. In addition to the salaries and wages paid by the companiesengaged in the mining and marketing of asbestos and asbestos products, the asbestos industry's morethan 300 suppliers of goods and services receive payments of around ZW$600 million each year,

WT/DS135/RPage 226

including over ZW$150 million for the state-owned Zimbabwe Electricity Supply Authority (ZESA)and the National Railways of Zimbabwe.

4.78 It is apparent from the foregoing that a ban on asbestos would have severe repercussions forthe Zimbabwean economy. In fact, as has been demonstrated, the ban on asbestos by France hasalready impacted negatively on the Zimbabwean economy. It must be mentioned in this connectionthat Zimbabwe views with great concern the potentially wider implications of the French ban on theuse of chrysotile asbestos. While it is true that most countries, including the United States, still do notgenerally prohibit the use of chrysotile asbestos or products containing chrysotile asbestos, there is theprobability that other governments may be tempted to follow the French example if the Frenchmeasure were upheld by the WTO. Indeed, the European Union has just announced - withoutawaiting the outcome of a WTO ruling - that it will move to ban the use of chrysotile asbestos in all itsmember States.133 Zimbabwe wished the WTO to be aware of the wider implications of the decision itwould render in this dispute.

4.79 Zimbabwe argues that the risks involved in the use of chrysotile asbestos can be adequatelycontrolled. It appears that the concerns that governments have with respect to the use of chrysotileasbestos relate to airborne asbestos dust or respirable asbestos fibres, as they may have an effect onhuman health. For this reason, the United Nations Environment Programme (UNEP), theInternational Labour Organization (ILO), and the World Health Organization (WHO), within theFramework of an Inter-Organization Programme for the Sound Management of Chemicals,commissioned a Task Group of international experts to make an evaluation of the risks for humanhealth from exposure to chrysotile asbestos and to make recommendations for health protection andfurther research. The report of the Task Group was published in 1998.134 One of the Task Group'smain conclusions was that "[e]xposure to chrysotile asbestos poses increased risks for asbestosis, lungcancer and mesothelioma in a dose-dependent manner".135 The Group acknowledged, however, that itwas not possible to provide quantitative estimates of risks to humans given the dearth of informationand data.136 Furthermore, the Group cautioned that there was a need for further epidemiologicalstudies of populations exposed to pure chrysotile so as to clearly and reliably be able to distinguishbetween chrysotile and amphibole exposure.137 In other words, there is a possibility that the availabledata may actually overestimate the risks to humans from exposure to chrysotile asbestos.138 What isquite clear from the Task Group's conclusion - and this is crucial - is that the risks to humans areconditional on exposure as well as on doses or concentrations.139 The key objective for anyresponsible government must therefore be to reduce exposure. This said, it should be borne in mindthat chrysotile asbestos is a natural product. It is present in the air we breathe and in the water we

133This information is based on a report by Reuters News Agency, dated 6 May 1999.134WHO, IPCS Environmental Health Criteria 203 - Chrysotile Asbestos, Geneva, 1998.135Ibid., p. 144.136Ibid., pp. 7 and 144. Zimbabwe notes that, in this regard, it is misleading for the EC to cite the 1998

Task Group report for the proposition that there is an international consensus that no threshold of exposure canbe identified below which no risk to humans exists. The Task Group in fact merely stated that it could notidentify any such threshold on the basis of the available data. See WHO, IPCS Environmental HealthCriteria 203 - Chrysotile Asbestos, Geneva, 1998, pp. 7 and 144.

137Ibid., p. 145.138Zimbabwe notes that this is especially true of applications of chrysotile-containing products in

industries such as construction, on which the EC has placed great emphasis in its Submission, because "studieshave not been in general able to distinguish between chrysotile and amphibole exposure". See WHO, IPCSEnvironmental Health Criteria 203 - Chrysotile Asbestos, Geneva, 1998, pp. 122 and 112.

139According to Zimbabwe, the EC confirms this when stating: "[L]es principales données qui ont étéprésentées illustrent le caractère ubiquitaire de l'amiante en milieu de travail qui peut, à des niveaux d'expositionsuffisamment élévés, entraîner de nombreux cas de maladies mortelles".

WT/DS135/RPage 227

drink. Exposure is therefore inevitable, and no ban can change that.140 With these facts at hand, thequestion arises as to whether the French ban of asbestos is justifiable given the information within thepublic domain. Zimbabwe believes that what is at the heart of this dispute is the risk of occupationalexposure to cement containing chrysotile asbestos. This is because prior to 1997, i.e. before theFrench ban was implemented, around 90 per cent of French imports of chrysotile asbestos fibres wereused for the production of asbestos-cement.

4.80 It is the submission of Zimbabwe that transportation and storage of imported chrysotileasbestos fibre do not entail a risk of exposure provided that there is proper packaging. Anotherpossible activity involving a risk of exposure is the production of chrysotile asbestos-cement itself. InZimbabwe, this risk has been contained, as demonstrated by the monitoring done by a group ofindependent experts for Turnall Fibre Cement Company Limited, which is a Zimbabwean companyengaged in the manufacture of chrysotile asbestos-cement. The focus of the research has been on thehealth hazards related to asbestos during the process of manufacture. This research has been going onfor more than ten years and so far there have been no reported cases of risks to human life. It shouldbe mentioned here that the EC has adduced as relevant evidence a very recent study by theU.K. Health and Safety Commission which is alleged to demonstrate that notwithstanding theapplication of control measures, "primary users" of chrysotile asbestos fibres, i.e. workers in asbestos-cement factories, showed a higher mortality rate in relation to asbestos-related lung cancer andmesothelioma. Zimbabwe views this study with considerable scepticism in view of the fact that thereare long latency periods involved in the above-named diseases and that the current "cases" go a longway back to a time when the control measures implemented were far less sophisticated than they arenow.

4.81 Zimbabwe asserts that a risk of exposure may also be incurred by workers or any person, forthat matter, during installation, maintenance and repair of asbestos-containing products. The risksinvolved in the use of asbestos-containing products can be adequately controlled, even taking intoaccount France's high level of protection against health risks, thus making a ban unnecessary.141 Infact, the 1998 Task Group report supports this conclusion when stating that "[n]on-friable productsand appropriate technological controls greatly reduce fibre release.142" It can thus be said that the riskof occupational exposure (i) is a function of the nature of the product and (ii) the risk inherent in thatproduct can in any event be further reduced through appropriate control measures. Regarding theproducts at issue, i.e. products made from asbestos-cement, the first thing that should be noted is thatasbestos-cement does not contain friable asbestos. Moreover, and equally importantly, products madefrom asbestos-cement are products of high density and thus chrysotile asbestos fibres are firmlyblended into the final product. This reduces to a minimum the likelihood of fibres being released intothe air and thereby posing a health hazard to human beings. The ILO came to the same conclusion ina report released in 1985: "[l]a manipulation de produits contentant de l'amiante dans lesquels lesfibres d'amiante sont solidement fixées dans un liant de telle sorte qu'il ne puisse pas se former depoussières ne présente pas de danger pour la santé.143"

4.82 It emerges therefore that when products made from asbestos-cement are used and handledproperly, the risks associated with their use are minimal. The recommendation of the 1998Task Force was to the same effect. It recommended that appropriate control measures be

140WHO, IPCS Environmental Health Criteria 203 - Chrysotile Asbestos, Geneva, 1998, pp 2 and 129

et seq.141For a more detailed discussion of this point, see Zimbabwe's arguments with respect to GATT

Article XX .142WHO, IPCS Environmental Health Criteria 203 - Chrysotile Asbestos, Geneva, 1998, p. 28.143Bureau International du Travail, La sécurité dans l'utilisation de l'amiante, Conférence internationale

du Travail, Rapport VI (1), 71ème session, 1985, Genève, p. 29.

WT/DS135/RPage 228

implemented wherever occupational exposure might occur.144 Among the control measures whichmight be used to minimize exposure to chrysotile asbestos are engineering controls, special workpractices (including workplace hygiene), and protective equipment, such as technical applianceswhich eliminate or minimize the formation of asbestos dust, as well as protective respiratoryequipment or special protective clothing. That risk control is in fact an effective means of dealingwith chrysotile asbestos-related health concerns is borne out by the following passage taken from thereport of the 1998 Task Group: "[d]ata from industries where control technologies have been appliedhave demonstrated the feasibility of controlling exposure to levels generally below 0.5 fibres/ml.Personal protective equipment can further reduce individual exposure where engineering controls andwork practices prove insufficient."145 In light of the foregoing, it is the contention of Zimbabwe thatthe combined use of high-density products made from asbestos-cement, which inherently are low-riskproducts, coupled with adequate risk control measures minimize the risk of exposure to asbestos dust.Whatever residual risk may remain does not, in Zimbabwe's view, justify an outright ban on chrysotileasbestos.

3. Legal Aspects

4.83 It is the submission of Zimbabwe that the French ban of chrysotile asbestos is contrary toWTO rules and should be lifted without any delay. It is the view of Zimbabwe that the French Decreeconstitutes a technical regulation within the meaning of the Agreement on Technical Barriers to Trade.As such, it must be in accordance with Article 2.2 of the TBT Agreement and hence must not be"more trade-restrictive than necessary to fulfil a legitimate objective". By totally banning the importof chrysotile asbestos, the French legislation contravenes the express language of this Article.Furthermore, in the event of the French Decree being found to fall outside the ambit of theTBT Agreement, the Decree contravenes the provisions of GATT Article III:4, as it discriminatesagainst imported asbestos in favour of other like products which are used in France for the samepurpose. In the same vein, the French Decree cannot be justified under the terms ofGATT Article XX(b), as claimed by the EC.

(a) The Agreement on Technical Barriers to Trade

4.84 Zimbabwe disagrees with the view of the EC that the Decree does not fall within the scope ofthe TBT Agreement. For a mandatory measure to come within the scope of the TBT Agreement, itmust be a "technical regulation". The Decree clearly is a mandatory measure. Notwithstanding theEC's claim to the contrary, it is the submission of Zimbabwe that the Decree, to the extent that itapplies to products containing chrysotile asbestos, qualifies as a technical regulation within themeaning of Annex 1 of the TBT Agreement. The argument of the EC that for the TBT Agreement tobe applicable, the Decree should have specified which particular products were covered by the ban iswithout any merit. It is the view of Zimbabwe that such an interpretation is overly restrictive.Annex 1 of the TBT Agreement talks about "product characteristics" in general. Nowhere does it statethat the national legislator should adopt only product-specific regulations. Even ignoring this point,Zimbabwe fails to understand why a Member should be precluded from laying down horizontal rulesapplicable to a group or groups of products which call for the same regulatory approach. In fact, itappears that there would be little merit in forcing Members to specifically enumerate all productscovered by a particular regulation when it is in the nature of things that new products would regularlyhave to be added to the list due to, for example, technological developments. From a public policyperspective, this would seem to be a rather inefficient and costly approach to adopt.

4.85 Zimbabwe argues that the second reason advanced by the EC in support of its argument thatthe TBT Agreement is not applicable in this case is also without any merit. According to the EC, the

144WHO, IPCS Environmental Health Criteria 203 - Chrysotile Asbestos, Geneva, 1998, p. 144.145Ibid., p. 144.

WT/DS135/RPage 229

ordinary meaning of the noun "characteristic" supports the view that, for the TBT Agreement to beapplicable, product characteristics must be positively defined. Applying this reading of theTBT Agreement to the present case, the EC argues that "not containing chrysotile asbestos" should notbe seen as the equivalent of a product characteristic. Zimbabwe finds this reasoning of the EC verytenuous. According to the Shorter Oxford English Dictionary, the noun "characteristic" designates a"distinguishing quality or peculiarity".146 Zimbabwe believes that, without doing injustice to theseterms, a product's "distinguishing quality or peculiarity" can lie in the fact that it does not containasbestos. The absence of any trace of asbestos clearly sets apart a product in terms of its qualitiesfrom another product which contains asbestos.147 In any event, Annex 1 does not actually requirepositive product characteristics. Zimbabwe submits that its interpretation of Annex 1 is also inconformity with the relevant context of Annex 1 of the TBT Agreement. All the Agreements annexedto the WTO Agreement are part of the relevant context.148 Thus, Article 2(f) of the Agreement onRules of Origin obliges Members to ensure that "their rules of origin are based on a positive standard".From this it follows that where Members wanted to give a special meaning to a term - in this case, tothe term "standard" - they used appropriate language to reflect their intention. Members did not adoptthat approach as far as Annex 1 of the TBT Agreement is concerned.149

4.86 Given the object and purpose of Annex 1 of the TBT Agreement, Zimbabwe wonders whatwould be the rationale of a rule which compels Members to define product characteristics positivelywhen all they care about is a negative characteristic. Why, for example, should France have topositively define the characteristics of a host of products when its only regulatory concern is with theasbestos contained in those products? It is the submission of Zimbabwe that its interpretation is alsoin conformity with the jurisprudence of the WTO Appellate Body. Thus, according to theAppellate Body, the term "measure" as it appears in various WTO agreements is to be understood toinclude a government's failure to act.150 In other words, a "negative" measure, i.e. a failure to act,counts as a measure no less than a "positive" measure. By token of the same reasoning, the term"characteristics" should encompass negative characteristics. In view of the above reasons, Zimbabwejoins Canada in believing that the general term "product characteristics" lends itself to aninterpretation which includes negative characteristics.

4.87 Having demonstrated that the Decree qualifies as a technical regulation under theTBT Agreement to the extent that it bans products containing chrysotile asbestos, Zimbabwe nowturns to show that the same is true also with respect to the Decree's ban on the use of chrysotileasbestos fibres as such. The EC has expressed the view that the French ban on the production andimportation of chrysotile asbestos fibre is not a technical regulation within the meaning of Annex 1 ofthe TBT Agreement because, just like the ban on asbestos-containing products, the ban on asbestosfibres is general (rather than specific) and lays down negative characteristics (rather than positiveones) or, for that matter, does not lay down any characteristics. As the issues of specificity and of"positive vs. negative standards" have already been discussed, the following submissions will focus onwhether or not the French Decree lays down product characteristics with regard to the ban on asbestosfibres. Zimbabwe contends that the matter is more complex than the EC makes it out to be. To besure, an independent and isolated ban on sales, say, of all cigarettes would not normally be considered

146The Shorter Oxford English Dictionary on Historical Principles, Oxford, 1993.147In this connection, Zimbabwe notes that it is a truism that it is often easier to define objects

negatively than to come up with an exact and exhaustive positive definition.148Panel Report, Japan – Measures Affecting Agricultural Products, adopted on 19 March 1999,

WT/DS76/R, para. 8.111.149"A treaty interpreter is not entitled to assume that [the use of different words in different places] was

merely inadvertent on the part of the Members who negotiated and wrote that Agreement". See EEC - MeasuresConcerning Meat and Meat Products (Hormones), Appellate Body Report, adopted on 13 February 1998,WT/DS26/AB/R, WT/DS48/AB/R, para. 164.

150See Guatemala – Anti-Dumping Investigation Regarding Portland Cement From Mexico,Appellate Body Report, adopted on 25 November 1998, WT/DS60/AB/R, footnote 47.

WT/DS135/RPage 230

a technical regulation. Yet the situation as it presents itself in this dispute is quite unlike that. AsCanada rightly pointed out, unlike cigarettes, asbestos fibres per se, i.e. as products in their own right,serve no useful purpose. It is the products containing asbestos fibres which have commercial use andvalue. By necessary implication, when it comes to dealing with the health hazards of asbestos, theconcern of policymakers and the law should be with products containing asbestos fibres, not withasbestos fibres, per se. If the products containing asbestos disappear, so will asbestos fibres.

4.88 Zimbabwe considers that the Decree is fully consistent with this straightforward principle.The EC does not contest this. On the contrary, the EC sets out the objective of the Decree as follows:"[l]'interdiction de l'amiante, en France et dans d'autres pays, n'a pas pour objectif de supprimer lesquelques 0,0002 fibres/ml qui existent 'naturellement' dans l'air. L'interdiction vise simplement àprotéger l'ensemble des travailleurs et des utilisateurs de l'amiante qui sont souvent exposés à desvaleurs très supérieures [...] pour des opérations courantes d'intervention sur des matériaux contenantde l'amiante-ciment."151 The EC explains the rationale of its asbestos-control policy in the followingterms: "[l]a politique adoptée en France en 1996 vise en tout premier lieu au remplacement desmatériaux contenant de l'amiante par d'autres matériaux sans danger [...]".152 It clearly emerges fromthese two quotes that the Decree aims at asbestos-containing products, not at asbestos fibres per se.The inference that can be drawn from this is that the import ban - just like the corresponding ban ondomestic production - does not perform an independent function, but a subsidiary one. Indeed, the ECexpressly states that nothing would change if the import ban - and, by implication, the ban ondomestic production - were lifted. Imported and domestically produced asbestos fibres could still notbe sold on French territory - because no products containing them could be sold. The followingsentence pinpoints this underlying logic of the French ban: "[l]e but est donc bien d'arrêter ladiffusion d'amiante le plus en amont possible".153 The ban on asbestos fibres is thus based onconsiderations of administrative efficiency, which is arguably only a secondary objective pursued byFrance. Again, this is confirmed by the EC: "[l]'interdiction d'importation a simplement pour but derendre plus efficace, en termes de contrôle, l'interdiction d'utilisation [which is France's primarygoal]".154

4.89 Zimbabwe asserts that, for the foregoing reasons, it should be readily apparent that the ban onchrysotile asbestos fibre is very closely related to the ban on asbestos-containing products. Assumingthe ban on chrysotile asbestos fibres could be viewed in isolation, it could possibly be argued that itdoes not, stricto sensu, lay down product characteristics. As Zimbabwe has demonstrated, however,such a line of reasoning is unwarranted and misses the point. The ban on asbestos fibres is an integralpart of the Decree. In fact, it is part and parcel of the same Article of the same Decree. Zimbabwetherefore submits that for purposes of this proceeding there is one single, indivisible regulatorypackage - the Decree - whose consistency the Panel needs to examine with the TBT Agreement.Zimbabwe is of the view that the Decree falls within the ambit of the TBT Agreement. This view isbuttressed by the reasoning of another Panel which faced a comparable situation. In the Kodak/Fujifilm case, the Panel had to decide whether a measure that had not been directly brought up underArticle 4 of the DSU could nevertheless be within the Panel's terms of reference. The Panel foundthat such a measure was not within the Panel's terms of reference, unless it was "subsidiary" or"closely related" to the measure that was properly before the Panel.155 By way of analogousreasoning, Zimbabwe argues that the French ban on asbestos fibres is "subsidiary" and "so closelyrelated" to the ban on asbestos-containing products - which, as shown, qualifies as a technical

151See Section III.B of this Report.152Ibid.153See Section III.C of this Report.154Ibid.155Japan - Measures Affecting Consumer Photographic Film and Paper, Panel Report, adopted on

22 April 1998, WT/DS44/R, para. 10.8.

WT/DS135/RPage 231

regulation within the meaning of the TBT Agreement - that it can reasonably be found to form anintegral part of the latter, and thus constitute a technical regulation in, and of, itself.156

4.90 Zimbabwe further submits that treating the ban on asbestos fibres and the ban on asbestos-containing products as separate and "unrelated" could give rise to unreasonable results. Such asituation could in fact arise in the present case. It could be envisaged, for instance, that the ban onasbestos fibres might be found to be consistent with the provisions of the GATT, while the ban onasbestos-containing products might be found to violate the provisions of the TBT Agreement because- to use but one example - it is more trade-restrictive than necessary to fulfil a legitimate governmentalobjective. Zimbabwe submits that such an outcome would be unreasonable and could undermine thepractical effectiveness of the TBT Agreement. Taken to its logical conclusion, such a situation wouldimply, on the one hand, that France could not produce asbestos-containing products domestically as aresult of the ban on imported or domestically-produced asbestos fibres. On the other hand, Francewould be required to lift its ban on imports of asbestos-containing products and adopt instead a lesstrade-restrictive measure which, in practice, would mean that a certain quantity of asbestos-containingproducts would cross the border into French territory. France would thus have no choice but to idly sitand watch as other countries take advantage of the business opportunities offered by the Frenchdomestic market. Zimbabwe is of the view that the drafters of the TBT Agreement did not and couldnot have intended such a result.

4.91 It is therefore the submission of Zimbabwe that the TBT Agreement applies to theFrench Decree in its entirety, i.e. with regard to the ban on asbestos-containing products as well as theban on asbestos fibres. The French legislation does not meet the requirements of Article 2.2 of theTBT Agreement, as amply demonstrated by Canada. Zimbabwe adopts the arguments presented byCanada in this connection and would like to support the views expressed therein by also relying on thearguments presented below on whether or not the French measure is necessary within the meaning ofGATT Article XX(b).

(b) The General Agreement on Tariffs and Trade

(i) Article III of the GATT

4.92 Zimbabwe argues that, in the alternative, and in addition to the claimed violations of theTBT Agreement, the Decree violates GATT Article III:4. Zimbabwe submits that chrysotile asbestosfibres and, at a minimum, cellulose fibres, aramid fibres and glass fibres are "like products" within themeaning of Article III:4. The EC confirms that cellulose and aramid fibres count among those fibreswhich are most frequently used to substitute asbestos fibres in the manufacture of cement.157

Cellulose, aramid and glass fibres are all produced in France.158 Whereas they may lawfully be sold inthat country, the importation and sale of asbestos fibres is prohibited. There is thus no doubt that

156Zimbabwe believes that its interpretation of the TBT Agreement also finds support in Article 1.6 of

the TBT Agreement. This Article states that "[a]ll references in this Agreement to technical regulations […]shall be construed to include any amendments thereto and any additions to the rules or the product coveragethereof, except amendments and additions thereof, except amendments and additions of an insignificant nature".This provision clearly establishes that the term "technical regulation" as used in the TBT Agreement is not to begiven a narrow reading, but one which promotes the Agreement's effectiveness.

157Zimbabwe notes that, with regard to glass fibres, this follows from the definition of HS tariffposition 68.11.

158In Zimbabwe’s view it does not matter for the purposes of an inquiry under Article III:4 whetherdomestic production of the "like product" is substantial or small. Nowhere does Article III:4 lay down arequirement that domestic production needs to be substantial.

WT/DS135/RPage 232

asbestos fibres are accorded "less favourable treatment" than cellulose, aramid and glass fibres,despite the fact that they are "like products".159

4.93 Zimbabwe notes that the EC contests that asbestos fibres, cellulose, aramid and glass fibresare "like products" within the meaning of Article III:4. It is well established in WTO jurisprudencethat the determination of whether or not products are "like products" must be made in accordance withsuch criteria as the products' physical characteristics and the products' end-use.160 It is equally clearfrom WTO jurisprudence that any such determination can only be made on a case-by-case basis, i.e.taking into account the specific and unique circumstances of each case.161 Regarding the firstcriterion, i.e. physical characteristics and properties, the EC claims that cellulose, aramid and glassfibres are not sufficiently similar to asbestos fibres in that their chemical composition is different. Inthis connection, Zimbabwe wishes to recall that the EC has acknowledged that the chemicalcomposition of all varieties of asbestos fibres is different as well. This did not preclude the EC,however, from concluding that chrysotile asbestos fibres and amphibole asbestos fibres were "likeproducts". Zimbabwe submits that the same logic applies and extends to cellulose, aramid and glassfibres.

4.94 Even ignoring the inconsistency in the reasoning of the EC, Zimbabwe does not believe thatthe differences pointed out by the EC are significant enough to make the relevant products "unlike"within the meaning of Article III:4. Zimbabwe wishes to recall, first of all, that "likeness" does notrequire that products be "identical in all respects".162 The second thing that should be noted is that thesignificance that is attached to differences in physical characteristics depends on the particularcircumstances of each case. In this case, as has previously been stated, the starting-point of anyanalysis must be the fact that chrysotile asbestos fibres, as products in their own right, serve no usefulpurpose.163 Chrysotile asbestos fibres are predominantly used as "inputs" in the manufacture offibre-cement products. It follows that substitute fibres like cellulose, aramid or glass fibres, on the onehand, and asbestos fibres, on the other hand, should not be compared to each other as products in theirown right. Instead, asbestos fibres and the relevant substitute fibres should be compared to each otheras products incorporated into cement. It is obvious that if this approach is adopted, as it should be, thedifferences identified by the EC become minor ones and irrelevant. The EC essentially makes thepoint that cellulose and aramid fibres are, on average, less fibrillose and larger in diameter thanasbestos fibres and that only asbestos fibres are internationally recognized as "category I" products,i.e. as products that have been shown to cause cancer. With regard to these alleged varying degrees ofhealth risk associated with the fibres at issue, it should be noted that whatever differences existbetween the relevant products become far less relevant when the fibres are blended with othermaterials to produce cement and other related products.164 As explained by Zimbabwe, any remainingrisks arise from improper handling and manipulation of cement-products and not from the cement-products themselves. Beyond that, Zimbabwe is not convinced that much significance should be

159Zimbabwe considers that the Decree falls obviously within the scope of Article III:4 inasmuch as it is

a regulation which affects the internal sale of asbestos fibres.160Japan – Taxes on Alcoholic Beverages, Appellate Body Report, adopted on 1 November 1996,

WT/DS/8/AB/R, WT/DS10/AB/R, WT/DS11/AB/R, p. 20.161Ibid., p. 20.162Japan – Taxes on Alcoholic Beverages, Report of the Panel, adopted on 1 November 1996,

WT/DS/8/R, WT/DS10/R, WT/DS11/R, para. 6.21163This is precisely why, in the view of Zimbabwe, the different tariff classification of chrysotile

asbestos fibres, on the one hand, and of cellulose and aramid fibres, on the other hand, cannot provide any usefulguidance for purposes of determining "likeness" in this case. The Appellate Body has in fact confirmed in itsreport on Japan – Taxes on Alcoholic Beverages, adopted on 1 November 1996, WT/DS/8/AB/R,WT/DS10/AB/R, WT/DS11/AB/R, p. 22, that the value of tariff classification as a criterion for establishing"likeness" must be assessed on a case-by-case basis.

164According to Zimbabwe, it should be borne in mind in this context that diameter and fibrillosity arein any event relevant only to the extent that these characteristics correlate with health risks to humans.

WT/DS135/RPage 233

attached to the fact that only asbestos fibres are listed by the WHO as a "category I" product. In fact,even the EC concedes that there is a lingering uncertainty about the risks involved in the use ofalternative fibres. Zimbabwe submits that the fact that there are to date no known negative effects onhuman health from the use of alternative fibres does not necessarily mean that they are risk-free.165

Zimbabwe notes that the EC shares that view, for it expressly acknowledges that "… un risqueindétectable n'est pas égal à une absence de risque".166

4.95 With regard to the second criterion, i.e. commonality of end-uses, Zimbabwe submits thatasbestos, cellulose, aramid and glass fibres serve "substantially identical end-uses".167 Their chemicalresistance and reinforcing capabilities make them almost perfect substitutes for asbestos fibres.It is therefore not the case that chrysotile asbestos fibres are unique products, as the EC would havethe Panel believe. As previously noted, the EC, in fact, acknowledges that cellulose and aramid fibresare commonly used substitutes for asbestos fibres.168 Moreover, like Canada, Zimbabwe believes thatthe structure of the Decree is at least suggestive of the substitutability of asbestos fibres with otherfibres. This becomes clear if the French Decree is seen in terms of the functioning of the politicalprocess. If very close substitutes had not been available to the principal users of asbestos fibres at thetime the Decree was signed into law, it is reasonable to assume that they would have lobbied theFrench Government and in all likelihood would have secured a broader exception (allowing thecontinued use of asbestos fibres) than the one that is now in the Decree.169 In light of the foregoingconsiderations Zimbabwe believes that asbestos fibres and cellulose, aramid and glass fibres should beregarded as "like products" within the meaning of GATT Article III:4.

(ii) Article XX of the GATT

4.96 Zimbabwe argues that the Decree is not justified under paragraph (b) of Article XX because itis not "necessary to protect human […] health".170 More particularly, the Decree does not satisfy thenecessity requirement. GATT 1947 case law has established that a measure qualifies as "necessary"within the meaning of Article XX if there is "no alternative measure consistent with theGeneral Agreement, or less inconsistent with it, which [a Member] could reasonably be expected toemploy to achieve its […] policy objectives".171 Zimbabwe believes that it is sufficient for it toestablish that - even assuming that asbestos fibres posed more of a health risk to humans - there areless trade-restrictive measures available to France to achieve its health objective. The EC claims thatin order for France to achieve its health policy objective there was no measure reasonably available toit other than an outright ban on chrysotile asbestos fibres. In particular, the EC submits that controlmeasures used to minimize exposure to chrysotile asbestos fibres are not sufficient to ensure thatFrance reaches its high level of protection. It also argues that control measures are impracticable inthe case of the large group of "secondary users" of asbestos fibres, i.e. those workers anddo-it-yourself people who, in the absence of control measures, may be exposed to chrysotile asbestosdust during installation, maintenance and repair of products containing chrysotile asbestos. The

165Zimbabwe notes that this is all the more true in view of the fact that the kind of diseases at issue here

involve long latency periods.166See Section III.B of this Report.167United States – Taxes on Petroleum and Certain Imported Substances, adopted on 17 June 1987,

BISD 34S/136, para. 5.1.1.168According to Zimbabwe, the same is true for glass fibres.169Zimbabwe notes that the fact that there is an additional and special temporary exemption in Article 7

of the Decree for certain used and agricultural vehicles precisely suggests that no equivalent and affordablesubstitutes existed at the time the Decree was signed into law and that the sectors concerned successfully lobbiedthe Government to provide for a temporary exemption.

170Zimbabwe notes that the question to be answered by the Panel here is whether it was necessary forFrance to discriminate between asbestos fibres and "like" domestic fibres in order to protect human health.

171Thailand - Restrictions on Importation of and Internal Taxes on Cigarettes, Panel Report adoptedon 7 November 1990, BISD 37S/200, para. 75.

WT/DS135/RPage 234

problem is compounded, according to the EC, by the fact that in many instances "secondary users" donot have any information as to whether they are dealing with products that contain asbestos. The ECsubmits that even if they were given that information, control measures are costly and turn what wouldotherwise be a simple operation into a costly, complicated and awkward one. Furthermore, the ECbelieves that "une fois mis sur le marché, il n'existe plus aucun moyen raisonnable de contrôler l'usagede l'amiante et, en particulier, de contrôler des opérations banales (découpage, sciage …) que denombreuses personnes peuvent être amenées à réaliser".172

4.97 Zimbabwe is not convinced by the arguments of the EC. First of all, regarding theeffectiveness of control measures, Zimbabwe believes that the observance of certain work practicesand the use of technical appliances in accordance with the ISO standard 7337, for example, would besufficient to meet the maximum exposure level acceptable to France. The EC argues that, even wherespecial technical equipment is used when high-risk activities are undertaken, peak exposure levels toasbestos would still exceed the French maximum level. What the EC fails to mention, however, isthat, as argued by Canada, the wearing of protective respiratory equipment and humidification of thematerials during those activities could significantly reduce the exposure - so much so, in fact, that therespect of the French maximum level of exposure would be ensured. Regarding the argument of theEC that mandatory control measures are impracticable because they are too costly, Zimbabwe conteststhe relevance of such considerations. After all, whether or not these costs are too high, is a matter tobe left to the dictates of the market. If the producers of asbestos-cement face insufficient demand fortheir products because of expensive control measures imposed on their customers, they will go out ofbusiness or diversify into the production of cement using alternative fibres. Likewise, Zimbabwe doesnot see any merit in the argument that control measures make certain work procedures complicatedand awkward. Where certain practices are imposed by law, the question of whether they areappreciated by those who must follow them becomes meaningless.173 It certainly does not in itselfprovide a rationale for trade-restrictive measures.

4.98 While Zimbabwe recognizes that it may not be readily apparent to an inexperienced personwhether or not he/she is handling a product containing asbestos fibres, it is by no means justificationfor instituting a far-reaching ban on products which might contain asbestos fibres. It is the contentionof Zimbabwe that it would be possible under the WTO legal framework for Members to impose adisclosure requirement, which would enable purchasers to make informed decisions as to whether ornot they purchase products containing asbestos fibres. Where the materials have already beeninstalled or incorporated, say, in a building, Zimbabwe does not see why there could not be, forinstance, an asbestos warning message next to the evacuation instructions on a notice board of thatbuilding. Moreover and specifically with respect to the work of plumbers, electricians and the like,Zimbabwe does not see why the owner of an installation or building could not be required to makeavailable some sort of map which would document in which parts of the installation asbestos ispresent.174 With reference to the concern of the EC that the use of asbestos-containing productscannot sufficiently be controlled, especially when it comes to "secondary users" of such products,Zimbabwe again does not think that banning all imports of such products would solve the problem. Infact, it would raise more problems than it would solve. To begin with, if indeed the FrenchGovernment is so concerned about do-it-yourself users of asbestos-containing products, it could have

172See Section III.C of this Report.173Zimbabwe notes that, by way of analogy, it might be added here that many people think that the

wearing of seatbelts in cars makes driving more "complicated" and awkward. Yet many countries made thewearing of seatbelt a legal requirement.

174In this connection, Zimbabwe recalls that a tenant or house owner who wants to drill a hole in a wallto hang up a painting, for instance, also needs to know exactly where the electrical wiring and installations are,lest he/she wants to put his/her life at risk.

WT/DS135/RPage 235

easily banned the sale of such products in all do-it-yourself outlets.175 Furthermore, as a supportingmeasure, it could have also restricted the handling of asbestos-made products to certified experts, thuseliminating contact with asbestos by inexperienced people.176 The protection of workers, such aselectricians and plumbers, could also have been ensured relatively easily. The French Governmentcould have, for example, required certification, which would only be bestowed upon an individualonce he/she had successfully followed information and training courses on the use and handling ofasbestos-containing products. The French Government could also have laid out the precise workpractices and technical appliances that must be used in all contacts with asbestos-containing products.To ensure compliance, the regulations could authorize the imposition of heavy fines or a custodialsentence in the event of a wilful disregard of the government's regulations. Needless to say, it is alsoopen to a Member to run information campaigns, so as to raise awareness among workers of the risksof asbestos fibres and the procedures to be observed in all contacts with such fibres. It is clear fromthe foregoing that the French Government had a number of alternative measures at its disposal whichwould have interfered less with trade and at the same time would have assisted in realizing itsoverriding objective of protecting the health and safety of its citizens.

____________________

175Zimbabwe notes that, after all, in most countries, drugs cannot be bought in supermarkets, but only in

pharmacies upon production of a doctor's prescription.176To give another analogous example, Zimbabwe notes that in many countries the installation of

ceiling lamps and other electrical appliances may only be carried out by certified electricians.

WT/DS135/RPage 236

V. PANEL'S CONSULTATION WITH SCIENTIFIC EXPERTS

A. DETERMINATION OF THE PROCEDURE

5.1 The Panel noted that the dispute before it raised scientific and technical issues. At the firstsubstantive meeting, the Panel informed the parties of its intention to seek the opinion of individualscientific experts except where, in the light of the parties' written rebuttals, it concluded that such aprocedure was not necessary. The areas in which the Panel wished to obtain information included thecircumstances of exposure to chrysotile asbestos and the associated risks, as well as the effectivenessof the controlled use of chrysotile. The Panel invited the parties to submit their comments to it inwriting, particularly regarding the areas on which the experts were to be consulted, the possibleapproaches to such a procedure and the international or other bodies that could usefully be consultedin order to identify suitable experts.

5.2 In a letter to the Panel dated 14 June 1999, Canada proposed, in regard to the possibleapproaches to a procedure for consultation with individual experts, that five requirements should bemet, each one intended to ensure observance of the right of the parties to be heard at all stages of theprocedure: (i) the Panel should consult the parties on the choice of scientific experts; (ii) the Panelshould seek the opinion of the parties concerning the formulation of the questions to be put to theexperts; (iii) the Panel should provide the parties with an opportunity to make written comments on adraft report by each of the experts; (iv) the parties should be able to question each of the experts onthe content of his final report at a meeting with the Panel; (v) the parties should be given thepossibility to make written comments on the conclusions set out in the final report of each expert andtheir legal implications. Like the Panel, Canada also believes that the areas on which the scientificexperts should be consulted ought to include the circumstances of exposure to chrysotile asbestos andthe risks associated with present applications as well as risk management by the controlled use ofchrysotile asbestos. The experts should also be consulted in two other areas, namely the comparativetoxicity of the different types of asbestos fibres and substitute fibres, and risk assessment methods,including the question of whether there are exposure thresholds below which the risk is undetectablein practice. In Canada's opinion, there are four specializations that in one way or another cover theabove-mentioned areas and from which the experts should be drawn. These are toxicology,epidemiology, risk analysis and occupational health. Given the scientific characteristics of thedispute, Canada would wish that each question be submitted to more than one expert, and that eachexpert submit an individual report. As regards the international institutions that could usefully beapproached in order to identify suitable experts, Canada believes that they should be consulted inorder to come up with a sampling of experts in the above-mentioned domains. The main selectioncriteria and hence the best guarantee of impartiality should be that experts must have conductedrecognized and independent research into chrysotile asbestos. The international organizations thatcould be approached included the World Health Organization, the International Labour Office and theInternational Organization for Standardization. Once a list of prospective candidates has been drawnup with the help of the international organizations, the parties should then be able to submit their ownlist of names of specialists who could act as scientific experts in the areas mentioned above.

5.3 In a letter dated 14 June 1999, the European Communities were of the opinion that thescientific issues raised in this dispute were simple and clear. The DSU rules on the burden of proofalso provided the Panel with sufficient guidance in dealing with the factual and scientific issues raisedby the parties to the dispute. With respect to the general selection procedures and criteria, theEuropean Communities believed that the Panel's use of experts for obtaining scientific and technicaladvice should respect the general principles of law. In particular, it should be transparent, avoidconflicts of interest, reinforce the integrity of the dispute settlement mechanism and foster publicconfidence in the outcome of the dispute. In the view of the European Communities, the Panel can inthis case establish only an expert review group under the terms of Appendix 4 to the Dispute

WT/DS135/RPage 237

Settlement Understanding. Indeed, the measure at issue in the present dispute is one that must beexamined strictly in terms of the GATT 1994, to the exclusion of the Agreement on Technical Barriersto Trade. Article 13:2 of the DSU provides as follows: "… With respect to a factual issue concerninga scientific or other technical matter raised by a party to a dispute, a Panel may request an advisoryreport in writing from an expert review group. Rules for the establishment of such a group and itsprocedures are set forth in Appendix 4". The establishment of an expert review group is the onlyoption provided under the DSU for panels wishing to obtain information on scientific matters. Thefirst sentence of Article 13:2 applies only to situations in which a Panel wishes to obtain factual ortechnical but not scientific information. In their context, the ordinary meaning of the terms, as well asthe object and purpose of Article 13:2, first and second sentences, clearly lead to the conclusion thatpanels are not authorized to deviate from the procedure laid down by Appendix 4 to the DisputeSettlement Understanding. Whether the request comes from a party or arises at the initiative of thePanel itself makes no difference. Strictly scientific matters cannot be resolved by means and/orprocedures other than those envisaged in Appendix 4 to the Dispute Settlement Understanding. Thechapeau to Appendix 4 to the Dispute Settlement Understanding also confirms this interpretation byproviding that the rules and procedures set forth in the Appendix "… shall apply to expert reviewgroups established in accordance with the provisions of paragraph 2 of Article 13", that is, regardlessof whether the Panel bases itself on the first or second sentence of that Article. This interpretation issupported by the fact that, if the Agreement on Technical Barriers to Trade (TBT) should beapplicable (which is not the case), Article 14:2 of that Agreement explicitly prescribes that panelsestablish only a technical expert group (which is equivalent to an expert review group). In such acase, the procedural rules set forth in Annex 2 to the TBT Agreement must apply. Annex 2 to thislatter agreement and Appendix 4 to the Dispute Settlement Understanding are almost identical.Moreover, by virtue of Article 1:2 and Appendix 2 to the DSU, only Article 14:2 of the TBTAgreement is applicable.

5.4 The European Communities also point out that the previous cases in which panels requestedthe opinion of scientific experts all came under the Agreement on the application of Sanitary andPhytosanitary Measures, which is not applicable in this case. Those previous cases are thereforeirrelevant to the present dispute. The dispute concerning Shrimp is the only other case for which theopinion of scientific experts was requested under the GATT 1994. But this example per se is notenough to set a valid precedent applicable to all cases, especially because the parties to the Shrimpdispute apparently did not request the exclusive application of Appendix 4 to the Dispute SettlementUnderstanding. The result is that, in the present case, should the Panel decide to seek the scientificopinion of external experts, it can do so only under Article 13:2, second sentence of the DisputeSettlement Understanding or under Article 14:2 of the TBT Agreement.

5.5 According to the European Communities, Appendix 4 to the DSU and/or Annex 2 to theTBT Agreement lay down almost identical rules on the establishment of an expert review group.These rules must all be observed in this dispute. Moreover, to ensure that the aforementionedprinciples are respected, the European Communities believe that the Panel should observe thefollowing specific criteria when choosing scientific experts: (i) the experts should not be citizens ofthe parties to the Dispute; (ii) the Panel should select scientific experts in different areas ofspecialization in order to ensure coverage of all the areas identified by it. These areas are: the humanhealth hazards posed by asbestos, especially chrysotile asbestos; the inapplicability of a threshold;the circumstances of exposure and the question as to whether what is known as "controlled use" caneliminate the potential hazards to human health; (iii) the European Communities believe that if thePanel decides to request information, it should consult at least five experts so that more than oneexpert will have the requisite expertise and provide answers to the questions in the various areasidentified by the Panel. In the light of the number of experts that the Panel should consult, onlyscientists with proven expertise in the realm of asbestos should be selected; (iv) the experts should bedrawn mainly if not exclusively from the International Agency for Research on Cancer (IARC), aspecialized agency of the WHO. The IARC has studied asbestos from all possible angles and should

WT/DS135/RPage 238

therefore be well placed to propose experts covering all the areas in which questions could be posed.The Panel should also explore the possibility of consulting the International Labour Office (ILO) inthe event that the IARC is unable to cover all the areas in question; (v) the experts chosen must haveno link whatsoever, present or past, with the industry producing asbestos or substitute products. Theymust furthermore clearly demonstrate the lack of any conflict of interest. The parties should receive atthe outset the Curricula Vitae of all the candidates proposed and should have at least ten working daysin which to verify the skills, expertise and possible conflicts of interest of the candidates; (vi) thePanel should also request the opinion of the parties as to the aim of the consultation with experts, thetype and nature of the questions to be put to them; (vii) the aim of the consultation should be tofurther the knowledge of the scientific considerations germane to this dispute. Therefore, and inaccordance with the provisions of the Dispute Settlement Understanding, the questions to be put bythe Panel must have a direct and strict bearing only on the scientific aspects of the case. The questionsmay not relate to legal problems nor to any problem of interpretation of any WTO Agreement underexamination.

5.6 Having taken cognizance of the comments from the parties, the Panel decided to consult theexperts on an individual basis, pursuant to paragraphs 1 and 2, first sentence, of Article 13 of theUnderstanding on Rules and Procedures Governing the Settlement of Disputes. The Panel convenedthe parties to a meeting on 10 July 1999 to acquaint them with the procedure it intended to follow andto give them the opportunity to state their opinions on the matter. The Panel recalled Article 13 of theDispute Settlement Understanding which, among other things, provides that:

"Each panel shall have the right to seek information and technical advice from any individual ortechnical body which it deems appropriate." [ … ]

"Panels may seek information from any relevant source and may consult experts to obtain their opinionon certain aspects of the matter."

5.7 At that meeting, the Panel told the parties that, in its opinion, Article 13 of the DisputeSettlement Understanding empowered it to seek such information and technical advice as it deemed fitin a given matter; in particular, a panel was free to determine whether it was necessary or appropriateto establish an expert review group. In the case at hand, the consultation of experts acting in their ownright seemed to it to be the most appropriate form of consultation. The Panel intended to seekinformation concerning the circumstances of chrysotile exposure and the attendant hazards. In thecircumstances, the Panel indicated that it would structure its questions around the following maintopics: the pathogenicity of chrysotile, the relative pathogenicity of amphiboles, chrysotile andsubstitute products; the assessment and management of risks associated with the use of chrysotile;the effectiveness of controlled use of chrysotile.

5.8 The Panel then presented to the parties the procedure that it intended to follow, which is thesame used by previous panels that had consulted experts selected on an individual basis:

• The experts will be placed under the authority of the Panel. They will be consulted ona personal basis and not as representatives of a government or organization. Theiropinion will be strictly in the nature of advice; it will not be binding on the Panel;

• the number of experts to be chosen by the Panel will be decided depending on thenumber of matters on which an opinion will be sought, as well as the number ofmatters on which each expert can give an opinion;

• the Panel intends to request names from the World Health Organization (WHO), theInternational Labour Organization (ILO), the International Programme on Chemical

WT/DS135/RPage 239

Safety (IPCS), the International Agency for Research on Cancer (IARC), theInternational Organization for Standardization (ISO), and from the parties;

• the Panel does not intend to appoint experts who are citizens of one or other of theparties to the dispute, unless the parties consent to their appointment or the Panelbelieves that it would otherwise be impossible for it to secure the specializedscientific advice needed;

• the Secretariat will request the persons suggested to submit a curriculum vitae. Thecurricula vitae will be transmitted to the parties. The parties may not establish contactwith the experts suggested;

• the parties will have an opportunity to make comments and to state any majorobjections they may have to any expert under consideration. The Panel will informthe parties of the experts it chooses;

• the experts will receive all the relevant elements of the communications on aconfidential basis;

• the Panel will prepare draft questions for the experts. They will be communicated tothe parties. The parties will have the opportunity to comment on the questionsproposed or to suggest additional questions before they are sent to the experts. ThePanel will then draw up a definitive list of questions which will be sent to the expertsand simultaneously to the parties;

• each expert will receive all the questions. He will be requested to reply to thequestions falling within his sphere of competence and, if necessary, to indicate theareas on which he does not feel competent to reply. The experts will be invited toprovide written answers; copies of those answers will be transmitted to the parties.The parties will have an opportunity to make written comments on the replies fromthe experts and the replies will be included in the Panel's final report;

• should the Panel deem it fitting, either on its own initiative or at the request of a party,a meeting may be held with the experts immediately before the second substantivemeeting. Before the meeting, the Panel will ensure that: (i) experts are made privy tothe parties' comments on their replies; (ii) the experts each receive the replies of theother experts to the Panel's questions;

• the minutes of the meeting with the experts will be submitted to the parties and to theexperts so that they may make corrections. The corrected version will be attached tothe Panel's final report.

5.9 The Panel gave the parties the opportunity to transmit their written comments to it.

5.10 In a letter dated 19 July 1999, Canada recalled all the points that it had notified to the Panelin its letter of 14 June 1999. Canada agrees with the Panel as to the nature of the information andadvice that it intends to seek from the scientific experts. It nevertheless believes that the experts bestqualified to reply to the Panel's questions concerning the circumstances of exposure to chrysotile andthe associated hazards are to be found in the areas of toxicology, epidemiology, risk assessment andoccupational safety. In addition to the opportunity given to the parties to make written comments onthe experts' replies, the Panel should also provide for the possibility of a final written submission bythe parties following the second substantive meeting. As regards the stipulation that the scientificexperts may not be citizens of any of the parties to the dispute, Canada believes that this procedural

WT/DS135/RPage 240

rule, established in Appendix 4 to the Dispute Settlement Understanding, normally applies only to theestablishment of an expert review group. In the Hormones case, the Appellate Body stated in thatconnection: " … once the Panel has decided to request the opinion of individual scientific experts,there [was] no legal obstacle to the Panel drawing up, in consultation with the parties to the dispute, adhoc rules for those particular proceedings".1 As the agreement of the two parties to the dispute isrequired if the selection of citizens of one of the parties is to be allowed, Canada is surprised at therefusal of the European Communities to allow the selection of their citizens. Canada is prepared toconsider the selection of experts who are citizens of the European Communities despite the refusal ofthe European Communities to consider experts from Canada. In this dispute, if the citizens of theparties are automatically excluded, the Panel risks facing a situation in which it will be unable toselect the experts with the best scientific knowledge considering the nature of the advice being sought.Canada therefore requests the European Communities and the Panel to reconsider their decision withregard to the non-participation of citizens of the parties.

5.11 Moreover, Canada cannot accept that, as demanded by the European Communities, theexperts must clearly demonstrate the absence of any conflict of interest. It is not incumbent upon aprospective expert to prove his impartiality, instead he is merely required to fill out a disclosure formconcerning his interests, relationships and any matters that may affect his independence. This form isprovided for in the document entitled Rules of Conduct for the Understanding on Rules andProcedures Governing the Settlement of Disputes.2 Once the persons approached as potential expertshave filled out their disclosure forms, the parties to the dispute may oppose any candidate who hasdisclosed an interest, relationship or matter that may place him in a situation of conflict of interest.The Panel is empowered to decide whether the information disclosed in the form really places acandidate expert in a situation of conflict of interest and to uphold a party's objection to an expert'scandidature. The approach taken by the Panel in the Shrimp case should be followed in this instance.Having noted that in their disclosure forms three of the experts approached had disclosed what mightbe considered as potential conflicts of interest, the Panel nevertheless decided to confirm theirappointments "being of the view that the disclosed information was not of such a nature as to preventthe individuals concerned from being impartial in providing the scientific information expected ofthem. The Panel has also taken into account the disclosed information when evaluating the answersprovided. The Panel underlined that, in making its choice, it had been guided primarily by the need togather expertise of the best quality and covering as wide a field as possible. In [the circumstancesspecific to this case], it was difficult – if not impossible – to reconcile this need with an agreement byall the parties to the dispute on each and every individual concerned".3 Canada is surprised at theEuropean Communities' insistence on the absence of any link between the experts and producers ofchrysotile asbestos but not between the experts and anti-asbestos pressure groups. No one opposesthe principles of independence and impartiality of experts or the observance of the rules on conflictsof interest. The single pertinent consideration remains the way in which these principles should beapplied in this particular instance.

5.12 In a letter dated 19 July 1999, the European Communities took note of the Panel's decisionto consult individual scientific experts pursuant to Article 13:1 of the Dispute SettlementUnderstanding. The European Communities contest the legal basis of the Panel's decision. Under theinternational customary principles of treaty interpretation, a systematic interpretation of Articles 13:1and 13:2 of the Dispute Settlement Understanding suggests that as far as scientific matters areconcerned, the preferred option in the Dispute Settlement Understanding is the establishment of anexpert review group. The term "scientific matter" appears only in the second sentence of Article 13:2

1 EC Measures Concerning Meat and Meat Products (Hormones) Report of the Appellate Body,

WT/DS26/DS48/AB/R, adopted on 13 February 1998, para. 148.2 WT/DSB/RC/1, of 11 December 1996.3 United States – Import Prohibition of Certain Shrimp and Shrimp Products, Report of the Panel,

WT/DS58/R, adopted on 6 November 1998, para. 5.7.

WT/DS135/RPage 241

of the Dispute Settlement Understanding, which envisages only the constitution of an expert reviewgroup. The drafting history of the WTO Agreements also confirms this interpretation.4 Thethree previous cases in which panels sought the opinion of scientists in their own right all had to dowith matters arising under the SPS Agreement, Article 11:2 of which expressly mentions "scientific"matters and envisages the possibility of consulting experts individually.5 Canada furthermore requeststhat the TBT Agreement be applied to the measure at issue here. It is worth noting that Article 14.2 ofthe TBT Agreement provides only for the possibility of consulting a technical expert group. ThisAgreement contains no provision equivalent to Article 13:1 of the Dispute Settlement Understandingor to Article 11:2 (first sentence) of the SPS Agreement. The very terms of Article 14.2 of theTBT Agreement are therefore different from Articles 13:1 and 13:2 (first sentence) of the DisputeSettlement Understanding and from Article 11:2 of the SPS Agreement. This difference is notaccidental.6 It denotes the clear intention of the WTO Members to settle scientific or technical mattersin the framework of the TBT Agreement only by establishing an expert review group. The decision ofthe Panel to consult experts on a personal basis is also contrary to Article 1:2 of the DisputeSettlement Understanding, which provides as follows:

"To the extent that there is a difference between the rules and procedures of this Understanding and thespecial or additional rules and procedures set forth in Appendix 2, the special and additional rules andprocedures in Appendix 2 shall prevail."

5.13 As explained above, there is a clear difference between Article 13:1 and 13:2 (first sentence)of the DSU, invoked in this case by the Panel, and Article 14:2 of the TBT Agreement. The specialrules and procedures mentioned in Appendix 2 to the DSU, namely Article 14:2 of theTBT Agreement, which provides for the establishment of a technical expert group, should thus beapplied in the present case, should the Panel judge the TBT Agreement to be applicable.7 Therefore,the European Communities consider the Panel's decision contrary to the letter, object and purpose ofArticle 14:2 of the TBT Agreement (if the latter is applicable), in conjunction with Article 1:2 of theDSU, and to Article 13:2 (second sentence) of the DSU. Besides, from a systematic point of view,the Panel's decision renders useless and obsolete the provisions of the Dispute SettlementUnderstanding and of the TBT Agreement regarding expert review groups, which are clearly theoption preferred by WTO Members and the only one for which rules of procedure have been drawn upin the WTO for the settlement of "scientific" questions.8 At this stage, the European Communities aretherefore obliged to reserve all their rights on this issue. They would also request the Panel, inkeeping with current WTO practice and for the sake of transparency and due process, to state inwriting the criteria and the reasons for its decision to call on individual scientific experts and thereasons for which is has not entertained the arguments put forward by the European Communities, andto communicate this information to the parties to the dispute.

4 According to the European Communities, confirmation can also be found in the chapeau to

Appendix 4 to the Dispute Settlement Understanding which provides that "[the following rules and procedures]shall apply to expert review groups established in accordance with the provisions of paragraph 13", that is,regardless of whether it is the first or the second sentence of this Article that is being used by the Panel.

5 That also explains the Appellate Body's reasons for its finding on that matter in the Hormones case.See the report AB/1997-4, para. 147.

6 As the Appellate Body found in Hormones (para. 164), "a treaty interpreter is not entitled to assumethat such usage was merely inadvertent on the part of the Members who negotiated and wrote that Agreement".

7 The third sentence of Article 1:2 of the Dispute Settlement Understanding is not applicable in thiscase, as the GATT 1994 does not contain contradictory rules and procedures on this matter.

8 The Panel's interpretation is also contrary to one of the corollaries of the general rule of interpretationset out in the 1969 Vienna Convention, which is that the interpretation must give meaning and effect to all theterms of a treaty. As held by the Appellate Body in the Gasoline case "an interpreter is not free to adopt areading that would result in reducing whole clauses or paragraphs of a treaty to redundancy or inutility"(AB-1996-1, page 22). Specifically, the Panel has so far refrained from providing explicit substantive reasonsfor its choice of consultation with individual experts over the establishment of an expert review group.

WT/DS135/RPage 242

5.14 As regards the type of scientific background and specializations, the European Communitiestake the view that the experts should be cancer specialists, in particular in lung cancer andmesothelioma. They should also be epidemiologists experienced in the area of asbestos and cancer.The European Communities are not clear as to what type of scientific discipline would encompassthose persons who would be required to provide advice regarding "risk evaluation and management inthe use of chrysotile" and "the effectiveness of the controlled use of chrysotile", nor what type oftechnical expertise they should have. If such experts exist, they should be able to provide informationabout all the categories of persons who could come into contact with asbestos and asbestos-containingproducts, such as those working in maintenance, repair and construction (for example, carpenters,plumbers, heat repairers, workers in insulating materials, do-it-yourself enthusiasts, etc.). TheEuropean Communities believe that the scientists chosen should also have expertise in the inspectionof houses, buildings and factories for the presence and possible removal of asbestos. Obviously, suchexperts cannot be allowed to have any link, whether direct or indirect, with the industries producingasbestos or those producing the equipment for reducing the risk of asbestos fibre inhalation. Such alink would seem particularly possible if the experts were to be designated by the ISO. The EuropeanCommunities consider that at least two experts should be designated for each scientific domain andeach area of questions. That is a minimum prerequisite for a balanced view and for not being entirelydependent on the views of just one person. At all events, the overall number of experts should not beless than six.

5.15 The European Communities have expressed their wish to receive copies of the letters to besent by the Panel to the aforementioned institutions under this point and of their replies. The expertsappointed should not be nationals or residents of the parties to the dispute. The EuropeanCommunities consider that all the candidates must submit a detailed curriculum vitae in time so as toenable the parties to verify their scientific credentials, experience and independence. The candidatesmust therefore clearly indicate in their curriculum vitae whether in the course of their professional lifethey have worked for or provided advice, in whatever form, to the industries producing asbestos,asbestos-containing products and substitute products or to the industry producing "controlled use"equipment. In addition, the selected experts must complete a disclosure form concerning potentialconflict of interest, pursuant to the Rules of Conduct for the Understanding on Rules and ProceduresGoverning the Settlement of Disputes adopted (WT/AB/WP/3, Annex II, page 16, 28 February 1997).The disclosure form must contain all the information indicated in the illustrative list appearing inAnnex II to the Rules of Conduct mentioned above. It should also explicitly contain information as towhether the expert has done any type of paid or unpaid work (scientific research, consulting, expertadvice, participation in the board of directors or board of management, etc.) for the enterprisesengaged in the extraction, production, processing of or trade in asbestos, asbestos-containing productsor substitute products, or for enterprises producing the equipment intended for "controlled use".

5.16 It is the opinion of the European Communities, that the Panel should, for example, requestthat the disclosure form further indicate: (i) the expert's professional situation (job in an enterprise orinstitute connected to the asbestos, substitute products or "controlled use" equipment industries);(ii) whether the expert is a member of the board of directors, board of management or any othersupervisory body within an enterprise, association, institution or interest group linked with theindustries producing asbestos, substitute products or equipment for "controlled use"; (iii) whether hehas conducted scientific research or provided expert advice at the request of or under contract to anenterprise, association, institution or interest group connected with the industries producing asbestos,substitute products or a "controlled use" equipment.9 If the aforementioned clarifications and

9 As far as the European Communities are concerned, additional support for the proposition that the

term conflict of "interest" should be interpreted as broadly as possible may be drawn from Article III.1 of theRules of Conduct mentioned above, and in a systematic interpretation (by analogy) of the following provisions:

WT/DS135/RPage 243

information are not given in the curriculum vitae and in the disclosure form, the parties will not be ina position to exercise their rights and make the type of comments being requested of them by thePanel. Therefore, the European Communities consider that the issue of the scientific credentials,experience and, in particular, that of the independence and impartiality of the experts, are ofparamount importance and that they need to be reflected in the Panel's decision on the selection andconsultation of scientific experts. They therefore wish to reserve their rights until completion of theselection procedures. The parties should be allowed sufficient time to enable them to make effectivelyknown to the Panel their views on the above issues. Specifically, they should be given sufficient timeto make known their views on the list of potential experts to be chosen by the Panel and to submittheir comments on the written replies from the experts to the questions put to them by the Panel.

5.17 In a letter to the parties dated 2 August 1999, the Panel confirmed its intention to consultexperts individually, in application of Article 13 of the DSU. The Panel carefully examined thearguments advanced by the parties concerning the expert consultation procedures, in particular, theEuropean Community argument that Article 13.2 of the DSU Agreement requires the constitution of atechnical expert group as envisaged in Appendix 4 to the DSU for the purposes of consultation withexperts on scientific matters. Article 13 of the DSU provides, among other things, that "each Panelshall have the right to seek information and take advice from any individual or body which it deemsappropriate" and that "Panels may seek information from any relevant source and may consult expertsto obtain their opinion on certain aspects of the matter". In addition, Article 13.2 prescribes thatpanels "may" request an advisory report in writing from an expert review group specifically thoughnot exclusively to examine a factual issue concerning a scientific matter. The Panel deems this text toallow for the establishment of such an expert group, while not ruling out consultation of experts on anindividual basis, both with regard to a scientific matter "or other technical matter". This interpretationof Article 13:2 of the DSU seems to the Panel to be perfectly in line with the text of this provision,interpreted in accordance with Article 31 of the Vienna Convention on the Law of Treaties, and withthe interpretation given by the Appellate Body that Article 13 of the DSU does not prevent panelsfrom consulting with individual experts and leaves to the sound discretion of a panel the determinationof whether the establishment of an expert review group is necessary or appropriate.10

5.18 The Panel also considered the European Community's argument that, if the measure at issueshould be deemed to fall under the TBT Agreement, which the Communities contest, Article 14.2 ofthat Agreement would require the establishment of an expert review group for any scientific ortechnical matter, and the EC position that pursuant to Article 1:2 of the DSU, that provision wouldprevail over those of Article 13 to the DSU. Article 14:2 of the TBT Agreement is among theprovisions mentioned in Appendix 2 to the DSU and which, under Article 1:2 of that Understanding,will prevail over the provisions of the Understanding to the extent that there is a difference betweenthe two. The Panel notes, however, that it is only "to the extent that there is a difference" between therules and procedures of the Understanding and a special or additional rule or procedure in Appendix 2to the DSU that the latter will prevail. Yet, as stated by the Appellate Body, it is only where theprovisions of the DSU and the special or additional rules of Appendix 2 cannot be read ascomplementing each other that the special or additional provisions will prevail over those of the DSU, Articles 8:2, 8:3 and 17:3 of the DSU, paras. 2 and 3 of Appendix 4 to the DSU, as well as paras. 2 and 3 ofAnnex 2 to the TBT Agreement.

10 See the Reports of the Appellate Body in European Communities – Measures Concerning MeatProducts (Hormones) (WT/DS26/26-DS48/AB/R), para. 147 (" … in disputes involving scientific or technicalissues, neither Article 11.2 of the SPS Agreement, nor Article 13 of the DSU prevents panels from consultingwith individual experts. Rather, both the SPS Agreement and the DSU leave to the sound discretion of a panelthe determination of whether the establishment of an expert review group is necessary or appropriate") andArgentina – Measures Affecting Imports of Footwear, Textiles, Apparel and Other Items (WT/DS56/AB/R)para. 84 ("Article 13 of the DSU enables a panel to seek information and technical advice as it deemsappropriate in a particular case, and ( … ) the DSU leaves 'to the sound discretion of a panel the determination ofwhether the establishment of an expert review group is necessary or appropriate'.").

WT/DS135/RPage 244

that is, in a situation where the two provisions would be mutually incompatible.11 In the present case,Article 14:2 of the TBT Agreement provides that a panel "may" establish a technical expert group.Like Article 13:2 of the DSU, this text envisages the possibility of establishing a technical expertgroup and lays down the procedures that would be applicable in the event. Nevertheless, it does notexclusively prescribe the establishment of a technical expert group, and this possibility, in our opinion,is not incompatible with the general authorization given under Article 13 of the DSU to consult withindividual experts. The two provisions can be read as complementing each other.

5.19 The Panel believes that in this case the consultation of experts on an individual basis is themore appropriate form of consultation, inasmuch as it is the one that will better enable the panelusefully to gather opinions and information on the scientific or technical issues raised by this dispute.Considering in particular the range of areas of competence that might be required, it is appropriate inthis case to gather information and different individual opinions rather than asking for a collectivereport on the various scientific or technical matters in question. In the light of the foregoing, the Panelwishes to underline that its decision to consult experts on an individual basis is without prejudice tothe applicability of the TBT Agreement to the measure in question, on which the parties disagree.

B. SELECTION OF EXPERTS

5.20 The Panel has requested the assistance of five institutions in identifying experts. Theinstitutions concerned are the World Health Organization (WHO), the International LabourOrganization (ILO), the International Programme on Chemical Safety (IPCS), the InternationalAgency for Research on Cancer (IARC) and the International Organization for Standardization (ISO).The parties have also submitted names to the Panel. The Secretariat then requested those of theproposed experts who were prepared to participate to submit to it a detailed curriculum vitae. Thosecurricula vitae were forwarded to the parties, who were able to convey to the panel their commentsconcerning the potential experts and to indicate, where appropriate, whether they had any majorobjections to any of them. Upon careful examination of the curricula vitae and the comments of theparties, the Panel accepted the following four experts, whose nominations were not opposed by theparties:

• Dr. Nicholas H. de Klerk, Senior Research Fellow, Department of Public Health,University of Western Australia, Australia;

• Dr. Douglas W. Henderson, Professor of Pathology, Head of the Department ofAnatomical Pathology, Flinders Medical Center and The Flinders University of SouthAustralia, Australia;

• Dr. Peter F. Infante, Director, Office of Standards Review, Health StandardsProgramme, Occupational Safety and Health Administration, Washington D.C.,United States;

• Dr. Arthur W. Musk, Clinical Professor of Medicine and Public Health, University ofWestern Australia, and Physician, Department of Respiratory Medicine, Sir CharlesGairdner Hospital, Nedlands, Australia.

5.21 The experts were asked to acquaint themselves with the Rules of Conduct for theUnderstanding on Rules and Procedures Governing the Settlement of Disputes12, paying special

11 See the Report of the Appellate Body in Guatemala – Anti-Dumping Investigation Regarding

Portland Cement from Mexico (WT/DS60/AB/R), paras. 65 and 66.12 WT/AB/WP/3, of 28 February 1997.

WT/DS135/RPage 245

attention to Annex 2 (Illustrative list of information to be disclosed). No expert has disclosed anycircumstance that could be considered as the potential source of a conflict of interest.

5.22 In consultation with the parties, the Panel prepared precise questions which it submitted toeach expert individually. The experts were requested to answer only those questions that theyconsidered to be within their domain(s) of competence. Written communications by the parties,transcriptions of their oral statements, as well as the references they submitted to the Panel weretransmitted to the experts for their information. The written answers from the experts have beenforwarded to the parties, who have had a chance to comment on them. The questions posed by thePanel and the answers given by the experts are contained in section V.C. The observations of theparties are reproduced in section V.D.

5.23 On 17 January 2000, the experts were invited to discuss with the Panel and the parties theirwritten answers to the questions and to provide additional information. Annex VI to this reportcontains the minutes of the meeting.

C. QUESTIONS BY THE PANEL AND COMMENTS BY THE SCIENTIFIC EXPERTS

5.24 The Panel requested the experts to comment on the areas of difference between the partieshighlighted in the first paragraph of each question, as well as to address the specific points listed. ThePanel encouraged the experts to indicate, to the extent possible, key points on which they consideredthat (i) there is scientific proof, (ii) there is broad agreement among experts, (iii) there is uncertaintyand/or a range of divergent opinions among experts.

1. Introductory Comments by Dr. Henderson

(a) Introduction

5.25 This introduction sets out a general summary of prevailing knowledge and uncertainties onasbestos-related disorders, with emphasis on mesothelioma and lung cancer, together with discussionof both the amphiboles and commercial chrysotile, patterns of exposure, and some brief details ofin vivo and in vitro experimental studies.

5.26 This introduction has two purposes: (i) to provide a general background and broadperspective to the questions and answers that follow; and (ii) to correct some inaccuracies and errorsin the documentation supplied already to the WTO. In so doing, I have tried to broaden theperspective beyond the classical Canadian studies on the Quebec chrysotile miners and millers, andbeyond the INSERM Report. A number of the general discussions in this introduction have beentruncated after the issue has been put into context, and some of these discussions are then continuedand amplified in my specific responses to the questions. This has produced some iteration of somepoints, but I believe that the advantages - avoidance of the potential for distortion created by answerswithout adequate background information - outweigh any disadvantages. The division of my reportinto these sections also provides an opportunity to indicate the relative importance of epidemiologicalstudies versus in vivo or in vitro experimental models in the formulation of my opinions and answers.

5.27 At the outset, I emphasize that Australia (including Western Australia) is no longer anasbestos producer. Production of crocidolite at the Wittenoom blue asbestos industry stopped in 1966.None has been produced or exported since. Crocidolite was used in asbestos-cement products inAustralia until 1966 when its use was discontinued, but imported amosite was used in these productsuntil 1984 [NICNAS 99]13. The use of chrysotile in fibro-cement products was discontinued in 1987.

13For complete references, see Annex III to the Panel Report.

WT/DS135/RPage 246

5.28 As stated repeatedly in the documentation provided to the WTO, asbestos has the capacity toinduce at least five benign pleuropulmonary disorders, and two cancers: parietal pleural fibrousplaques; benign asbestos pleuritis with effusion; diffuse pleural fibrosis; rounded atelectasis;asbestosis; primary lung cancer; malignant mesothelioma. The essential characteristics of thesedisorders are discussed in the documentation submitted to the WTO and lie beyond the scope of thisreport; if further details are required, standard texts should be consulted [26-30]. There is nopersuasive or compelling evidence that asbestos of any type causes cancers other than lung cancer andmesothelioma, with the arguable exception of cancer of the larynx. At this stage, it is sufficient topoint out that: " ... there is an exposure-response relationship for all chrysotile-related diseases.Reduction of exposure through introduction of control measures should significantly reduce risks.Construction and demolition operations may present special control problems". [EHC 203, p 141].

(b) Malignant mesothelioma – Introduction and general observations on asbestos andmesothelioma

5.29 Malignant mesothelioma is a cancer of the mesothelial cells that line the serosal membranesof the major body cavities, namely the pleura, the peritoneum, the pericardium and the tunicavaginalis testis; the constituent neoplastic cells characteristically express the phenotype of arecognized pattern of differentiation of the mesothelium, whether epithelioid, sarcomatoid or both(biphasic), as revealed by conventional light microscopy, mucin immunohistochemistry,immunohistochemistry or electron microscopy, or a combination of these techniques [31-33]. Likeother forms of cancer, mesothelioma has the capacity for local invasion of tissues such as the chestwall or lung, with confluent serosal spread in most cases but not all, and in some instances distantmetastasis [31], with an almost invariably fatal outcome. Mesothelioma is resistant to conventionalcancer therapies (e.g. radiotherapy or chemotherapy), but some long-term survivals have beenrecorded following radical surgery (pleuropneumonectomy) in patients in good physical condition andwith early-stage disease [34-43]; radical surgery of this type is not a treatment option for the majorityof mesothelioma patients.

5.30 Most mesotheliomas encountered in the 1990s are a consequence of prior occupationalexposure to asbestos [24], including bystander exposure. The relationship between asbestos -especially one or more of the amphibole varieties - and mesothelioma is accepted by virtually allauthorities as causal. In this respect, asbestos fulfils all The Bradford Hill Criteria for theestablishment of causality (e.g. please see Stolley and Lasky [44]).

5.31 The following points about asbestos and mesothelioma are worth emphasis:

(i) Inhalation of asbestos fibres represents the overwhelming cause of mesothelioma inindustrialized societies, so much so that the incidence of mesothelioma is usually considered to be anindex of those societies' past usage of asbestos.

According to Peto et al. [24]:

"The great majority of mesotheliomas are caused by asbestos, and the much higher incidence in menindicates that most are due to occupational rather than environmental exposure. The incidencecontinues to rise approximately as the third power of time since first exposure to asbestos for manydecades after exposure has ceased (Peto et al., 1982), and most patients are men first exposed 30 ormore years ago. A country's mesothelioma rate is therefore a quantitative indicator of its population'spast exposure — mainly occupational — to asbestos." [p 666].

5.32 Boffetta [15] claims that:

WT/DS135/RPage 247

"Asbestos is the only established risk factor of mesothelioma. Because of the rarity of the disease andthe specificity of the causal association, all cases occurring among asbestos-exposed workers areattributed to this exposure." [p 476; please see following discussion].

4.33 The asbestos exposure may take the following forms: (i) direct or indirect occupationalexposure (including bystander exposure); (ii) domestic exposure: e.g. household contacts of asbestosworkers, such as wives who washed the asbestos-contaminated work clothes of their husbands [45-47]; (iii) environmental exposure: this category includes those who lived downwind of asbestosindustries or in townships contaminated by asbestos [45-47]. For example, ≥ 27 mesotheliomas havebeen recorded among those who lived at Wittenoom as children (the roads, airstrip and school yardswere surfaced with crocidolite tailings from the mine, and children often played in the mine tailings).

"The tumor [mesothelioma] is more often seen in workers who have only moderate or small amount ofasbestos in their lungs, and who show little, if any, clinical or radiologic evidence of pulmonaryfibrosis. This amount of asbestos may be inhaled not only by professional asbestos workers, but alsoby those who handle products containing only a small proportion of asbestos, those who do not handleasbestos at all but merely work alongside asbestos workers such as craftsmen employed in the buildingindustry — carpenters, electricians, etc. — those who have relatives who carry asbestos home in theirworkclothes and those who live close to asbestos plants." [47] [p 295].

5.34 No history of asbestos exposure is obtainable in about 15-25 per cent of mesothelioma cases[31, 48]. Nonetheless, absence of a history of exposure does not equate to absence of exposure, andevidence indicates that many of these mesotheliomas are in reality attributable to asbestos inhalation— e.g. remote, brief or forgotten exposure, or alternatively, the individual may be unaware that he (themale:female ratio is about 8:1) was in fact exposed to asbestos: (i) from my own series ofmesotheliomas, 79 per cent of the request forms that accompanied the biopsies on which the diagnosiswas made gave a positive history of past asbestos exposure; clinical review of the remaining 21 percent yielded a history of asbestos exposure in a substantial proportion, including some for whom theoriginal history stated that there was no exposure, so that my estimate of the proportion for whom apositive history of exposure was eventually obtained is ≥ 85-90 per cent. This estimate is inreasonable agreement with figures in the 1999 Report for the Australian Mesothelioma Register[AMR 99], where 85 per cent of mesotheliomas had a history of asbestos exposure; (ii) Leigh et al.[49] found measurable asbestos fibre levels (> 200,000 fibres per gram dry lung tissue) in 81 per centof the 28 per cent of Australian mesothelioma cases that had no history of occupational orenvironmental exposure to asbestos.

(ii) Putative or possible factors other than asbestos implicated in the induction of mesothelioma

5.35 In the reply to Question 3 from the European Communities, Canada makes the followingstatements:

"... Canada wishes to inform the European Communities of the considerable body of evidencecontradicting their statement that asbestos in all forms (amphiboles and chrysotile) is the only knownfactor that can cause mesothelioma or pleural cancer. ... A number of studies suggest other potentialrisk factors that may have been under-estimated in epidemiological studies in industrialized countries.... A number of artificial fibres cause mesothelioma when they are injected into the pleura andperitoneum of laboratory animals. Note also that the International Agency for Research on Cancer(IARC) has classified refractory ceramic fibres as probable carcinogens, partly because of instances ofmesothelioma induced by inhalation and injection in animal experiments. The SV40 virus readilyinduces mesothelioma when injected into animals; studies suggest that the virus contaminated anti-polio (poliomyelitis) vaccines from 1955 to about 1963 and may induce mesothelioma with or withoutthe help of asbestos fibres. Some studies of humans report the presence of the simian SV40 virus in thebiological tissue of mesothelioma victims. Ionizing radiation used in cancer therapy and perhapsoccupational exposure to radiation have induced mesothelioma. ... [E]rionite has been shown to beeven more toxic than crocidolite in causing mesothelioma; it has killed large numbers of villagers in

WT/DS135/RPage 248

Turkey. Erionite is a mineral fibre but does not belong to the asbestos family."

5.36 Possible factors other than asbestos implicated as contributory or causative for raremesotheliomas are tabulated below:

TABLE 1: PUTATIVE OR POSSIBLE RISK FACTORS AND MEDIATORS OF RISK OF MESOTHELIOMAOTHER THAN ASBESTOS

Factor Comments

Erionite Very high incidence of mesothelioma due to environmental exposure inTurkey (restricted geographic localization only).

Chronic inflammation Pleural scars (tuberculosis, pleurisy, therapeutic pneumothorax, familialMediterranean fever); see following discussion.

Radiation Single cases after Thorotrast injection or radiotherapy; causality unproven.One case in atomic bomb survivor.

Beryllium Two doubtful cases described.Vegetable fibres No proof in humans.Hereditary factors Familial cases (explicable by common asbestos exposure ± unidentified

genetic susceptibility factors, including association with other cancers infirst-degree relatives).

Immunological factors Rapidly progressive cases in patients with HIV infection; very rare —single case(s) only.

Dietary factors Provitamin A, β-carotene may decrease the risk (unproven).Viruses Mesotheliomas in animals. Simian virus 40 (SV40) DNA sequences

reported in mesotheliomas; see following discussion

Modified from Hillerdal [20].

5.37 There are anecdotal reports of mesothelioma following radiation, including radiotherapy forchildhood cancer such as Wilms' tumour [50-56]. In addition, excess rates of mesothelioma have beenreported among both Danish and German patients exposed to radioactive thorium dioxide (Thorotrast)for radiological procedures [57, 58], although a similar but smaller Japanese study found no suchexcess [59]. Neugut et al. [60] investigated women with breast cancer and patients with Hodgkin'sdisease, many of whom had been treated by radiotherapy (RT):

"The authors performed a retrospective cohort study utilizing 251,750 women registered with breastcarcinoma in the Surveillance, Epidemiology, and End Results Program of the U.S. National CancerInstitute from 1973-1993, 24.8% of whom received RT as part of their initial management, and 13,743people with Hodgkin's disease, 50.6% of whom received RT as part of their initial management.RESULTS: Six cases of malignant pleural mesothelioma were found: two in breast carcinoma patientstreated with RT and four found in women not treated with RT. No cases occurred in the patients withHodgkin's disease. The overall estimated relative risk for malignant pleural mesothelioma after RT was1.56 (95% confidence interval, 0.18-5.63). CONCLUSIONS: To the authors' knowledge, this is thefirst controlled study to investigate thoracic radiation exposure and malignant pleural mesothelioma,and no association was found." [abstract].

5.38 I am also aware of at least one mesothelioma in a patient with HIV infection (AIDS) [61].Other mesotheliomas have occurred many years after chronic inflammatory lesions of the pleura —e.g. chronic empyema or packing of the pleural cavity with leucite spheres as treatment fortuberculosis [62, 63], and there are a few reports of an association between familial Mediterraneanfever (FMF) and mesothelioma (about eight cases only; possibly related to recurrent FMF serositis[64-67]). However, cases of this type are exceptional and most cases of "post-inflammatory"mesothelioma with a short interval between inflammation and tumour (e.g. ≤ 2-3 years by analogywith the criteria for the diagnosis of benign asbestos pleuritis [33, 68, 69]), are probably

WT/DS135/RPage 249

mesotheliomas that presented with a burst of inflammatory activity, followed by a period ofquiescence [70].

5.39 In addition, background asbestos exposure represents a confounding factor for some casesassociated with radiation and immunodeficiency: (i) in one report on mortality among 260 plutoniumworkers, all six mesotheliomas occurred in individuals who had also sustained asbestos exposure [71]:"... no apparently elevated causes of death except for six cases of mesothelioma and six cases ofastrocytoma glioblastoma multiforme. The mesothelioma cases had a documented occupationalexposure to asbestos ..." [extract from abstract]; (ii) in one of my own cases, the patient had beentreated for Hodgkin's disease by mantle radiotherapy 10 years before the diagnosis of his primarypericardial mesothelioma, but he also had a background of occupational exposure to asbestos; (iii) inanother case — a pleural mesothelioma in a renal transplant recipient — the patient had also sustainedearlier occupational exposure to asbestos.

(iii) Erionite and mesothelioma in Turkey

5.40 Erionite (a fibrous zeolite) represents a naturally occurring fibrous mineral implicated in theinduction of mesothelioma in certain villages (notably Karain and Tuskoy) in the Cappadocian regionof Turkey [72, 7 3], and in Turkish emigrants [74]. So far as I am aware this represents a restrictedgeographic pocket of mesothelioma cases induced by erionite used as stucco or whitewash inbuildings, so that the inhabitants were exposed to high concentrations of erionite fibres from birth.Erionite has no relevance to the broader mesothelioma problem in Western Europe, North America,and Australia. Nonetheless, in its physical properties erionite has similarities to the amphibolevarieties of asbestos and it has been suggested that its greater mesotheliomagenicity is related to agreater surface area (200 m2 per gram) than crocidolite (8-10 m2 per gram), due to the presence ofpores in the crystal lattice (see Roggli and Brody [75]); such differences in surface topography mightcorrelate with differences in free radical generation at the surface of fibres.

(iv) Simian virus 40 (SV40) and mesothelioma

5.41 Recently, a voluminous literature has grown rapidly on the detection of SV40 DNA in upto 60 per cent of human mesotheliomas [76-87] and some other tumours, such as papillary carcinomaof the thyroid [88], osteosarcomas and brain tumours [83, 89-91]. These observations followed aninitial finding that SV40 could induce mesothelioma in hamsters when injected into the pleural cavity[92], and the later demonstration that SV40 could inactivate the tumour suppressor genes p53 and theretinoblastoma gene (Rb) via the large T antigen (TAG) [80, 82, 93, 94]. For humans, earlypoliomyelitis vaccines contaminated with SV40 were a potential source for the SV40 DNA [82-84].The following points on this interesting association are also worth emphasis:

• It has been suggested that the presence of SV40 might explain: (i) why mesotheliomaonly develops in a relatively small proportion of asbestos-exposed individuals(usually < 10 per cent); and (ii) why no history of asbestos exposure is obtainable ona sizeable minority of mesotheliomas [95]. However, almost all the mesotheliomas inwhich SV40 DNA has been found were asbestos-associated; to the best of myknowledge, there is no reported case-control analysis of SV40-associatedmesotheliomas where asbestos fibre counts were not elevated above reference values,with the exception of a recent study by Mayall et al. [96] (please see followingdiscussion). Therefore, the existing data do not adequately address either (i) or(ii): there are many other possible explanations for these observations.

• In other studies, SV40 or TAG could not be detected within mesotheliomatous tissue[97-99]. Galateau-Sallé et al. [100] found that SV40 was present not only inmesotheliomas but also in benign inflammatory disorders of the pleura and non-

WT/DS135/RPage 250

neoplastic lung tissue. In an as yet unpublished investigation carried out incollaboration with Prof. Alec Morley in the Department of Haematology-Oncology atthe Flinders University, we have also identified SV40 in mesotheliomas, and in non-neoplastic pleural lesions, normal tissues and colon cancers, casting doubt on thespecificity of the association.

• Two epidemiological studies have shown no increase in the incidence of bone or braintumours — or mesotheliomas — 30 years after the use of polio vaccines contaminatedwith SV40 [101, 102], although in a later study using SEER data14, Fisher et al. [103]reported an increased frequency of these tumours in subjects who had received SV40-contaminated poliomyelitis vaccines.

• The evidence so far only points to SV40 as a possible co-factor for asbestos in thegenesis of mesothelioma [96]. For example, Mayall et al. [96] detected SV40sequences in five of seven asbestos-associated mesotheliomas, but none of fourmesotheliomas that were not asbestos-related (investigated by fibre burden analysis oflung tissue, using electron microscopy). However, the evidence in favour of SV40 asa co-factor for mesothelioma induction is still inconclusive and non-persuasive, and inhumans the SV40 may represent an innocent bystander or passenger: the criteria forcausality [44] have not been fulfilled.

"It remains to be shown whether the presence of SV40 contributes significantly to malignanttransformation or whether certain human neoplasms provide a microenvironment that favorsviral replication in humans with latent SV40 infection." [91] [last sentence of abstract].

5.42 The point of these comments is that the evidence for a role of SV40 in the development ofmesothelioma is inconclusive, and most of SV40-associated cases still represent asbestos-associatedmesotheliomas. Although the literature contains anecdotal reports of mesothelioma followingradiation, some of these cases (e.g. among plutonium workers) are complicated by coexistent asbestosexposure and it is worth emphasizing that these cases are rare: together they add up to only a smallfraction of 1 per cent of the total burden of mesotheliomas in industrialized societies, for whichasbestos remains the overwhelming cause. As emphasized already, there is general agreement that theincidence of mesothelioma in various nations is a reflection of the past usage of asbestos by thosesocieties.

5.43 Hillerdal [20] comments along similar lines:

"... SV40 might be a cofactor to asbestos in some patients with mesothelioma, but the [findings] havenot been confirmed and are still disputed. ... In summary, then, as far as is known today, factors otherthan mineral fibres can only explain a very small proportion of mesotheliomas, and can for practicalpurposes be disregarded [i.e. when approaching the causation of mesothelioma among large cohorts orpopulations]. Thus, a malignant mesothelioma can be regarded either as caused by asbestos orbelonging to a normal background level — that is a spontaneously occurring tumour." [p 506].

(v) Male: female ratio for mesothelioma

5.44 Asbestos-induced mesothelioma affects males more often than females in a ratio of about 8:1,as a reflection of occupational exposure.

(vi) Anatomical distribution of mesothelioma

5.45 With the exception of one series in which 44 per cent of mesotheliomas were peritoneal [104],

14The National Cancer Institute's Surveillance Epidemiology and End Results program.

WT/DS135/RPage 251

there is general agreement that primary asbestos-induced mesothelioma affects the pleura more oftenthan the peritoneum, in a ratio of at least 3:1 or even up to ≥ 11:1 [31, 33] (see also AMR 99). InAustralia, ≥ 91 per cent of mesotheliomas arise in the pleural cavities, whereas about 7 per centrepresent primary peritoneal mesotheliomas and ≤ 1 per cent affect the pericardium or tunica vaginalistestis [33]. This predominance of pleural mesotheliomas in comparison to the peritoneum appears tocorrelate with gender differences in the frequency of occupational exposure to asbestos (the same highratio of pleural to peritoneal tumours is also encountered in the United States). In females, a smallerproportion of mesotheliomas arises in the pleura, and in one study of Swedish insulation workers, allseven mesotheliomas arose in the peritoneum [105] (please see following discussion).

5.46 One report [106] that included cases notified to the Australian Mesothelioma Registerfrom 1986 through 1988 gave figures for the anatomical sites affected in men and women: 676 of 723men had a pleural mesothelioma (93 per cent), whereas 38 were peritoneal tumours (5 per cent) andnine occurred in other sites (1 per cent). In contrast, 84 mesotheliomas in 101 women were pleural inlocation, whereas 17 per cent had a peritoneal mesothelioma.

5.47 Presumably, this difference in anatomical distribution between sexes is a reflection ofdifferent rates of occupational exposure to asbestos. On theoretical grounds, one would expectmesotheliomas entirely unrelated to asbestos to occur with about equal frequency in the pleura andperitoneum, or more often in the peritoneum because of the greater surface area of the peritonealcavity.

5.48 A partial list of the factors that might explain the higher proportion of peritonealmesotheliomas in some series and in women includes the following [33]:

• The high proportion of pleural mesotheliomas in men is presumably a reflection ofasbestos exposure, with deposition of asbestos fibres in lung tissue, followed bytranslocation of the fibres to the pleura; on this basis, asbestos inhalation appears toskew the proportional distribution of mesothelioma towards the pleura in comparisonto other sites. In contrast, fibres presumably follow a more circuitous route from thelung to the pleura, across the diaphragm and into the peritoneal cavity, to induceperitoneal mesothelioma; higher inhaled doses of asbestos might be necessary for therequisite number of fibres (whatever that is) to reach the peritoneum via the pleura, inorder to induce peritoneal mesothelioma.

• The high proportion of peritoneal tumours in some series may be a consequence ofpatterns of referral for cases that constitute problems in diagnosis, because thediagnosis of peritoneal mesothelioma is, in general, more difficult than for pleuralmesotheliomas. This may explain the higher proportion of peritoneal mesotheliomasamong cases referred to the US-Canadian Mesothelioma Panel [107], because manyof these represented problems in diagnosis, whereas the Australian MesotheliomaSurveillance Program (AMSP) captured all mesotheliomas throughout Australia [48].

• Genuine biological differences in the inhaled dose, deposition or transport of differentasbestos fibre types in some groups of workers, notably insulation workers [108] andformer Wittenoom workers [109] — as a consequence of heavy occupationalexposure — and in women [106, 110].

5.49 In answer to questions posed by the European Communities (Question 3, see Annex II), thefollowing comment is made:

WT/DS135/RPage 252

"... malignant diffuse mesothelioma is a cancer of the mesothelial cells of the pleura, the pericardiumand the peritoneum. Furthermore, peritoneal mesothelioma is an even more typical result of exposureto amphiboles than pleural mesothelioma."

5.50 From the preceding discussion on the proportions of mesotheliomas arising in the pleuralcavities versus the peritoneum, it is evident that this proposition is not correct: use of the term typicalin this context is inappropriate. In reality, pleural mesothelioma is a more typical or usual outcome ofasbestos exposure, whereas asbestos-induced peritoneal mesotheliomas are usually associated withmore prolonged and heavier exposures than pleural mesotheliomas, so that the proportion of patientswith asbestosis is higher than for pleural mesothelioma [111]. It has also been claimed that peritonealmesotheliomas are almost always a consequence of amphibole exposure (as opposed to chrysotileonly) [112]. Nonetheless, although some of the peritoneal mesotheliomas in my own series of casesfollowed high-dose exposures to asbestos that included one or more of the amphiboles, a few followedlower cumulative exposures, and Neumann et al. [111] have reported peritoneal mesotheliomas as aconsequence of exposure in the building trades and metal industries, in addition to asbestos industries;Rogers et al. [3] recorded peritoneal mesotheliomas in whom only chrysotile fibres were detected onlung fibre analysis (see Table 9, paragraph 5.137).

(vii) Latency intervals (lag-times)

5.51 In all reported studies, mesothelioma is a disease of long latency between exposure to asbestosand the subsequent diagnosis of the mesothelioma. In the AMSP [48], the mean latency interval (lag-time) was 37 years, with a reported range of 4-75 years; the lag-time was reported to be < 10 years inonly four of 499 asbestos-associated mesotheliomas (0.8 per cent). Many authorities set a minimumlag-time of 10 years (e.g. The Helsinki Criteria [113]), and for most patients the lag-time is in therange of 20-40 years, When the lag-time is < 10 years, it is likely that the proximate exposure wascoincidental, and that there were one or more earlier exposures.

(c) Spontaneous or background mesothelioma: does it exist?

4.52 The rare occurrence of mesothelioma in childhood and even as a congenital malignancysupports the existence of a background of spontaneous mesotheliomas unrelated to asbestos (inaddition, mesothelioma has been reported in fish (trout) [114], where inhalation of airborne asbestosfibres cannot be invoked). However, in epidemiological studies on adult populations, it is virtuallyimpossible to separate spontaneous mesotheliomas from those that are arguably attributable toenvironmental exposure to asbestos [70, 115]. The incidence of mesothelioma in women issometimes used as an index of the background or spontaneous rate: the crude incidence rate forwomen in Western Australia is about 2.6 per million person-years at age ≥ 15 years [115]. Theincidence rate in other populations is listed in Table 2 (following page).

5.53 In the answer to the WTO Panel's questions to Canada (Question 9, see Annex II), thefollowing statement is made:

"Recent analyses of Canadian data on mesothelioma in Canada, British Columbia and Quebec all agreethat the incidence rate of mesothelioma has been stable among women of all age groups since 1984.The rates are 70% higher in Quebec than in the rest of Canada, presumably as a result of more frequentand more intense exposure in the workplace."

5.54 The statistics for Australia differ on this point (Table 2): mathematical modelling ofWestern Australian data suggests that the incidence rate in women has risen about two-fold from the1970s until the 1980s, which might be explicable by increased general environmental exposure toasbestos, plus some occupational exposures among women [70, 115] (please see also AMR 99 —i.e. the graph for the Age Specific-Incidence Rates of Malignant Mesothelioma in Australia Women,1986-1995, especially for ages 50-64 and 65-79). This increased incidence among women

WT/DS135/RPage 253

presumably reflects direct or indirect occupational exposure, domestic exposure or environmentalexposure [115]; in this respect, it is worth emphasizing that domestic (household contact) exposure toasbestos — e.g. among wives laundering the dust-laden workclothes of an asbestos-exposed husband— is not necessarily low-level exposure, and analysis of the asbestos fibre content of the lungs in asmall number of such patients indicates that this type of exposure can approach occupational levels[116].

TABLE 2: INCIDENCE OF MORTALITY OF MESOTHELIOMA IN VARIOUS COUNTRIES AND AREASOVER TIME, 1960S TO 1994 (PER MILLION INHABITANTS PER YEAR)

Country or area Year Males Females

United States 1968-81 2.1 0.8North America 1972 2.8 0.7Texas 1976-80 5.8 2.1Selected cities, United States 1970s 4.4-11.1 1.2-3.8United States 1986 7-13 1-2Nantes-Saint-Nazaire, France 1956-74 5.2 0.2Nantes-Saint-Nazaire, France 1975-84 17.2 0.8Nantes-Saint-Nazaire, France 1985-92 19.4 4.0Great Britain 1968-71 8.4 2.3Great Britain 1972-76 12.6 2.8United Kingdom 1983 17.5 3.2Great Britain 1968-71 20.7 4.3Great Britain 1982-86 30.5 4.9Great Britain 1987-91 44.0 6.4Australia 1982-88 28.3 3.3Australia 1994 49.9 4.8Denmark 1978-80 14.7 7.0Barcelona, Spain 1983-90 8.3 4.7Finland 1990-94 10 2.9

Modified from Hillerdal [20].

5.55 The often-cited background or spontaneous rate of mesothelioma of 1-2 per million person-years [10, 117], has in part also been derived from backward extrapolation of the incidence rates inmen, to the point where the estimated incidence rates for men and women diverged from each other(i.e. linear extrapolation to the point where the sex ratio = 1:1) [117] . Hillerdal [20] suggests that thisincidence probably represents a high estimate and comments in the following terms:

"... there seems to be a small spontaneous basal or background incidence of the tumour [mesothelioma]... However, it is of course possible that some of these background cases might in fact be due tooccupational, domestic, or even environmental exposure, unknown to (or forgotten by) the patientsthemselves. ... There are authors who claim that the presumed background levels must be very low,and retrospective searches for the tumour in the medical literature reveal no convincing cases ofmesothelioma before 1946, although such negative evidence is of questionable value.15 McDonald and

15Mark and Yokoi [118] have called into question the existence of mesothelioma in the absence of

asbestos exposure, pointing out that the early descriptions of pleural tumours may have dealt with localizedfibrous tumours of the pleura (four of the five tumours reported by Klemperer and Rabin [119]) or secondarycarcinoma. Thus, mesothelioma might represent a new disease consequent upon the industrial use of asbestos(analogous to AIDS) and it may disappear upon withdrawal of the causative asbestos from the environment(analogous to smallpox). In support of this proposition, these authors cited the records of the MassachusettsGeneral Hospital, where no examples of mesothelioma were diagnosed before 1946, in contrast to 100 autopsycases thereafter, in a total of 47,000 autopsies. They also referred to the Henke-Lubarsch Handbuch der

WT/DS135/RPage 254

McDonald, in a recent review, estimated the background level to be 1-2/million per year; they came tothis figure by extrapolating backwards from epidemiological studies from various countries. ... It isnevertheless possible that there is a background level of mesothelioma, — that is, that the tumour canoccur even in the complete absence of asbestos (or erionite) fibres. However, the data reviewed hereindicate that if so, this background level must be very low — probably much < 1 case/millionpeople/year. This figure comes from studies of industrialized countries, where background exposure toasbestos is unavoidable. What the true figure is can only be guessed ... ". [p 507].

5.56 De Klerk [115] and Comin et al. [70] have commented that in the absence of specificexposure to asbestos, the final estimated rate for both men and women in Australia is 2.6 per millionperson-years — higher than the equivalent figure of 1.6 for Los Angeles [115]. This difference maylend some support to the proposition that general environmental exposure to asbestos may haveproduced an increase in the mesothelioma rate in Western Australia [115]. However, it is difficult orimpossible in general to draw firm conclusions from differences between different studies, because ofvariation in the accuracy of diagnosis and differences in the ways that data are collected.

5.57 In response to questions from the WTO Panel (Question 9, see Annex II) the Canadiandocument also observes that:

"The incidence of [mesothelioma] among men levelled off after 1984 in British Columbia ... and seemsto have levelled off in Quebec after 1990 ... Finally, analysis of Canadian rates between 1973 and 1992... estimates that the risk is four times greater for men born before 1940 than for men born between1951 and 1955. Those analyses therefore suggest that the incidence of mesothelioma has levelled off inCanada, is declining in British Colombia, and has levelled off in Quebec ..." .

5.58 In response to these observations, I emphasize the following: (i) the incidence ofmesothelioma among Australian males shows little evidence of levelling off, and has continued to riseuntil 1994-1995 and thereafter (please see Table 2 and the 1998 and 1999 Reports for the AustralianMesothelioma Register); (ii) from the recent report by Peto et al. [24], it is also evident that theincidence of mesothelioma in Western Europe continues to rise, with particular emphasis on malesborn between 1945 and 1950 who used asbestos-containing products in the 1960s and the 1970s (andthe early 1980s).

(d) Magnitude of the mesothelioma problem

5.59 Malignant mesothelioma continues to represent a major health problem within industrializedsocieties, and together with lung cancer it represents the most important occupational cancer amongso-called blue-collar workers [121-124].

5.60 It has been estimated that across Western Europe, North America and Australia (combinedpopulation ~ 800,000,000), around 10,000 mesotheliomas and 20,000 asbestos-induced lung cancersoccur annually, related mainly to occupational exposure (about one mesothelioma for every 200 tonsof asbestos produced, taking into account the prolonged lag-times) [125]. Steenland et al. [126]estimate that approximately 9000-10,000 men and 900-1900 women develop lung cancer each year in

speziellen pathologischen Anatomie und Histologie, wherein the four pages devoted to tumours of the pleura didnot specifically acknowledge the existence of mesothelioma; the Henke-Lubarsch authors concluded that manycases described in the literature as primary pleural neoplasms were cases of lung cancer with spread to thepleura. I find the evidence for Mark and Yokoi's proposition to be underwhelming and unconvincing. The casereported in the 1920 paper by Du Bray and Rosson [120] is, I believe, a clear example of a mesothelioma, as isthe fifth case of Klemperer and Rabin [119]. Failure to diagnose a tumour is hardly synonymous with its non-existence, and the pathological features of many tumours have been delineated in quite recent times.Pathological diagnoses follow prevailing evidence and fashions, and because the groundwork for modernconcepts of mesothelioma was laid down in 1931 by Klemperer and Rabin, it is hardly surprising that thediagnosis became more widespread only after this time.

WT/DS135/RPage 255

the United States because of past exposure to occupational carcinogens, and more than half of theselung cancers are related to asbestos (this overall estimate is considered probably to be conservative).Predictions of asbestos-related diseases in Australia (population ~ 18,000,000) indicate that about13,000 cases of mesothelioma (range 8000-20,000), about 40,000 cases of lung cancer (range 30,000-76,000) and 1000 cases of asbestosis are likely to occur between the years 1987 and 2020 [70, 127].

5.61 More recently, Peto et al. [24] have predicted that about 190,000 mesothelioma deaths arelikely to occur throughout Western Europe (Britain, France, Germany, Italy, the Netherlands andSwitzerland) over the next 35 years. If one adds in lung cancer at a ratio of one lung cancer for everymesothelioma death, this figure would rise to 380,000 deaths, and if the ratio of lung cancers tomesotheliomas is 2:1 the figure rises to 570,000 deaths.

5.62 Overall, asbestos may have caused approximately 5,000,000 deaths across industrializedsocieties so far. When future deaths in so-called developing nations are added, the final toll is almostcertain to be substantially higher, especially because occupational exposures in those countries arelikely to be heavier (e.g. China). Estimates of this magnitude are likely to engender alarm amongthose who set social policy. Even so, it is important that this problem, like others (e.g. atomic energy),is approached with common sense, rationality and prudence, taking into account population-based riskestimates: it would be irrational to swap one risk for another higher risk if the two risks were equallyserious.

(e) Some general observations on approaches to risk assessment on society and onepidemiological studies of asbestos-related cancers

5.63 The documentation supplied to the WTO includes estimates of risks from low-level exposureto chrysotile in proportion to various other risk factors in society: in fact, the relative risk ofmesothelioma from low-level exposure to asbestos in place is the focus of considerable controversy.Clearly, detailed analysis of this issue is beyond the scope of this report, but the excess risk ofmesothelioma from very low-level exposure to asbestos — e.g. simple occupancy of public buildingsor schoolrooms where average asbestos fibre concentrations are about < 0.001-0.02 fibres per litre —appears to be very slight: about ≤ 5.5 mesotheliomas per million lifetimes of 80 years, or < 1 caseper 10,000,000 per year.

5.64 The estimated mesothelioma risk for a 10-year exposure to low levels of airborne asbestos inschools (age start 7-8 years; fibre concentration 0.00065-0.001 fibre per ml16 is in the range 6.6-20per million lifetimes (0.0825-0.25 per million person-years). Estimates of this type are predicated onlinear dose-response models with no threshold, and these have been the subject of argument andcriticism. The occupational groups from which they were derived were exposed to mixtures ofasbestos types, but one might expect the risk to be even less or "undetectably low" in nations whereonly chrysotile was used. At this point, it is sufficient to emphasize that - even if one accepts for amoment the no-threshold linear dose-response relationship - calculations indicate that a singleasbestos fibre (the so-called "one fibre" hypothesis) would only have a 50-50 chance of producing asingle excess mesothelioma among all the humans who have ever inhabited Planet Earth.

5.65 These observations on mesothelioma risk estimates for very low-level exposure to asbestos inbuildings do not contradict the earlier suggestion in this report that Western Australian and Australianincreases in mesothelioma incidence among females were a possible consequence of generalenvironmental exposure to asbestos: the two-fold rise in incidence in Western Australia could be dueto environmental fibre levels higher than those recorded in public buildings elsewhere.

16Given in various publications as fibres/ml, fb/ml, f/ml, f/mL and fibres/cm3.

WT/DS135/RPage 256

5.66 If it exists, the risk of mesothelioma from very low-level environmental exposure to asbestosneeds to be considered in proportion to other risks of death in society. The lowest death-risk at anyage occurs in girls aged 4-14 years and is ~ 100 per million per year, but the risk in late teenageincreases by 300-400 per million per year — attributable largely to increased travel by motor vehicle.A 40-year-old man at risk of mesothelioma from low-level exposure to asbestos during childhood(excess mesothelioma risk < 1 per million person-years) has an annual risk of death from all causes ofabout 2000 per million. In 1990, de Klerk [128] had this to say on the subject:

"In the US the acceptable (or possible litigation-proof) lifetime risk seems to be one per million. TheFDA have this as a set policy; the EPA approximates to it, and other agencies seem to employ similarfigures. In the UK the Royal Society has set a higher range with an annual risk of one per millionconsidered negligible, with any form of control unjustified; one per 100,000 considered low ("very fewwould consider action necessary" — e.g. 16,000 km air or rail travel); one per 10,000 moderate (fewwould commit their own resources to reduce risk" — e.g. 16,000 km car driving, working as acoalminer); one per 1,000 high (e.g. age 30-39, 16,000 km motor cycling); and one per hundredunacceptable (e.g. age 55-59, smoking 20 cigarettes per day, heavy exposure to crocidolite)."

5.67 Within this context, one might ask what constitutes a negligible as opposed to an acceptable(or unacceptable) risk? In the context of the foregoing discussion, one might argue that a negligiblerisk is a statistical and scientific concept: a risk so slight that it does not require preventive orremedial steps in comparison to other risks in society (although some might argue about the dividingline between negligibility and unacceptability in this context). With acceptability or unacceptabilityother factors come into play: they include, for example, social, political and industrial considerations— and the likelihood of litigation over any situation wherein the theoretical or estimated risk iselevated to a background level, no matter how slightly. Accordingly, a risk, though slight or evennegligible, might still be considered unacceptable in legal or sociopolitical terms.

5.68 Others may dissent from the acceptability of the Royal Society approach to low, moderate andhigh risks discussed above. Conclusions about the acceptability or unacceptability of risk will alsovary according to the seriousness of the risk (e.g. the approach to a lethal risk such as mesotheliomawould be quite different from a factor that caused a large proportion of the population to sneeze onceor twice); these assessments will also vary according to the avoidability of the risk, the individualsmaking the assessment, and the question of informed consent by those at risk.

5.69 Furthermore, society abounds with inconsistencies and contradictions over the relativity ofvarious risks. For example, some societies that regulate or propose a ban on chrysotile asbestos makeextensive use of radioactive materials — e.g. in nuclear power stations and the production ofradioactive isotopes for medical purposes. Even so, the use of fissionable materials for these purposesmay be justified and justifiable within those societies, because: (i) the risk of morbidity or death fromwell-publicized mishaps at nuclear reactors is still substantially less than the risk of death fromalternative energy sources (e.g. higher mortality rates among coalminers); and (ii) because thematerials in question can be regulated and controlled so that they are accessible to only a smallfraction of society (i.e. workers who can be trained in the controlled use of radioactive substances);and (iii) nuclear power does not contribute significantly to air pollution or greenhouse gas emissionsin comparison to the burning of fossil fuels.

5.70 In addition, Nicholson [129] places the problem into the perspective of voluntary versusinvoluntary risks:

"Rather than compare asbestos risks with voluntary risks (smoking, school football) or risks that remainhigh despite expenditures of substantial public and private money (aircraft and highway accidents), it isworthwhile to compare them with other involuntary environmental risks that are controlled byregulatory agencies (pesticide exposures, drinking water contamination). In a review of regulatoryactions taken by the FDA ... and the EPA it was found that for estimated population risks exceeding one

WT/DS135/RPage 257

death/year, the individual lifetime risks were usually regulated if they exceed 1/1,000,000 for a lifetimeexposure. Only eight of 31 carcinogenic exposure circumstances that exceeded this level were notregulated. They involved saccharin, aflatoxin, formaldehyde and polycyclic organic matter ... " [p 81].

5.71 In fact, it is my view that over-reaction to the low risks produced by asbestos in place maylead to a greater risk — i.e. the carcinogenic risks imposed by asbestos removal programmes. Twomesotheliomas encountered in my own practice during 1999 occurred not in asbestos removalworkers, but in others who sustained bystander exposure as a result of this activity: (i) pleuralmesothelioma in a lecturer who had walked to and from her classroom at an Australian university eachday over a period of weeks, through a building where an asbestos insulation removal programme wasbeing carried out; (ii) pleural mesothelioma in a fireman who attended fires in buildings thatcontained asbestos-cement products and who participated in clean-up operations thereafter; aboutonce a month for some years, he also attended and examined buildings where fire alarms had beenactivated by high atmospheric concentrations of asbestos fibres produced by removal programmes.(In addition, a recent survey in Finland found "occasional high fibre concentrations even insidepersonal protectors during asbestos removal work" [130]).

5.72 Two other important points are worth emphasis. First, because they usually focus on specificcohorts or groups of workers, epidemiological studies may fail to identify a small but real risk,because of low statistical power. In this respect, the documents submitted to the WTO state that itmay be impossible to prove a negative (absence of risk), but one can also state that absence of proofdoes not constitute proof of absence. For example, a number of investigations have failed to identify astatistically significant increase in the relative risk (RR) of cancer among individuals with parietalpleural fibrous plaques. In an extensive review of asbestos and lung cancer, Henderson et al. [131]commented along the following lines in relation to pleural plaques and lung cancer:

"Nurminen and Tossavainen [132] also emphasized the issue of statistical power; they calculated theRR for plaque-associated lung cancer in the general population to be as low as 1.1, given the prevalenceof 4.6% among unlikely exposed and 13.0% among probably exposed men with an estimated twofoldrisk of lung cancer. Detection of this RR at a level of statistical significance would require a populationsample of about 300 000." [p 102].

5.73 In a discussion of the Hughes-Weill study [133] on radiological asbestosis and lung cancer inNew Orleans asbestos-cement factory workers — one of the three key investigations that proposed anobligate intermediary step of pulmonary fibrosis for the induction of lung cancer by asbestos —Henderson et al. [131] also commented in the same review:

" ... the number of lung cancer cases [in the Hughes-Weill investigation] was small. What number ofworkers would be required in such a study to detect an increase in risk of, say 1.4, 1.56 or 2.0, asopposed to the risk in workers with chest x-ray opacities? ... person-years of follow-up equivalentto 20-50 expected cases would be required to have any reasonable chance of detecting RRs of 1.4 to 1.6at a level of significance of 0.05. ... The power level for the actual sample of 420 ... to detect a risk of1.5 would be about 40%. That is, a true effect would be falsely declared 'non-significant' 60% of thetime. ... The low power of the Hughes-Weill study is exemplified by the fact that ... lung cancer riskwas not significantly associated with duration of employment or cumulative exposure (there was afairly restricted range of employment periods) and even the association of lung cancer with fibrosis wasonly marginally significant." [pp 93-94].

5.74 The point is that a low, non-significant or undetectable risk in a small cohort may nonethelesstranslate into a substantial body of disease when spread over a large population: e.g. an RR of 1.1representing an increase in risk of 10 per cent may require a population size of 300,000 to bedetectable at a level of statistical significance of 0.05, whereas this 10 per cent increase in a commondisease such as lung cancer may amount to a substantial burden of disease when spread across a

WT/DS135/RPage 258

population of, say, 1,000,000, 10,000,000, or 100,000,000. (Please see also later discussion onmesothelioma among brake mechanics: answer to Question 2.)

5.75 Another point is that a high frequency of a cancer such as mesothelioma in a small populationmay be overshadowed in absolute numbers by a lower occurrence rate for the same disorder spreadover a large population. For example, among non-smoking former Wittenoom workers, mesotheliomais now the most common cause of death [70] (in most cohorts exposed to amphibole asbestos, < 10per cent will develop mesothelioma). Nonetheless, mesotheliomas among the Wittenoom cohortcontribute only 5-6 per cent of the total burden of mesothelioma across the Australian population[AMR 99]. For example, the 1999 Report for the Register records 189 mesotheliomas among theformer Wittenoom population with only a single exposure to asbestos, in comparison to 187mesotheliomas among carpenters/joiners with only a single exposure to asbestos; the point is that thelower risk of mesothelioma from asbestos exposure among carpenters has produced almost the samenumber of cases, because carpenters among the Australian workforce constitute a much largeroccupational group than the entire Wittenoom cohort of about 6000.

5.76 This observation also applies to the numbers of mesotheliomas among chrysotile miners andmillers in Quebec, in proportion to other cases among the general population of Quebec. Bégin et al.[134] divided Quebec mesotheliomas into three groups, as shown in the following Table:

TABLE 3: MESOTHELIOMAS IN QUEBEC, 1967-1990

Group Type of asbestos exposure Number ofcases

Average age Average duration ofexposure

1 Chrysotile miners and millers,Thetford and Asbestos, Quebec

49 62 ± 8.1 yrs 30.5 ± 13.7 yrs

2 Manufacturing, industrial insulation,shipbuilding yards of Quebec

50 56.7 ± 8.6 yrs 21.4 ± 14 .5 yrs

3 General construction/ buildingmaintenance industries of Quebec

21 57.7 ± 7.2 yrs 27.7 ± 7.2 yrs

From Bégin et al. [134].

5.77 In this study, Bégin et al. [134] also commented that "the incidence of pleural mesothelioma inchrysotile miners and millers, although not as high as in the crocidolite workers, is well above theNorth American male rate". They also observed that "asbestos exposures in Group 3, althoughdifficult to quantify on the basis of the record, appear to be often very low intensity". Bégin et al. alsocommented in the following terms:

"The present study documents an increasing incidence of malignant mesothelioma in chrysotile minersand millers of the eastern townships of Québec, with 49 cases in the last 23 years and a rate of 2.5 casesper year in the last 10 years in the primary industry, as compared with a rate of 0.3 per year in the yearsprior to 1969 ... To put these rates into perspective, a comparison of the incidence for the combinedpopulation of the Asbestos and Thetford townships of Québec of some 40,000 adult males or themaximal estimated workforce of 10,00-15,000 men [sic; surely this is a typographic error in theoriginal, and it should be 10,000-15,000], 20 years ago and currently at risk, reveals that the incidenceof mesothelioma in the chrysotile mining townships of Québec would give an annual incidence rateof 62.5 cases per million per year for the 1980-1990 period, or in chrysotile miners and millers ofQuébec, would give an incidence rate of 150-250 cases per million per year for the 1980-1990 period.These values are well below the annual incidence rate of the crocidolite mining townships of SouthAfrica, estimated at 542 cases per million per year, and well above the rate for the North Americanpopulation, estimated at between 2.5 to 13 cases per year per million adult males for the 1970-1980period, and 14.1 cases per year per million adult males in 1984 and 15 cases per million for 1980 andprojected to increase for the 1990s. ...

WT/DS135/RPage 259

Thus, our observations add information of interest to the on-going debate regarding the relativecarcinogenicity of different types of asbestos fibers. Our data suggest that some of the cases ofmalignant mesothelioma in Québec chrysotile miners and millers may not be necessarily attributable toamphibole and could be chrysotile-induced. Lung tissue burden analyses, a better indication ofexposure than tumour tissue burden, will be done on these cases to further investigate this point. ....

Finally, our data strengthen the view that a substantial number of malignant mesothelioma cases have arelatively short asbestos exposure, particularly seen in Group 3. In our study, 25% of all cases are insuch a category" ... [pp 539-541].

5.78 In one of the documents submitted to the WTO, it is argued that evaluation of, and actions on,risks should be based on probability rather than mere possibility. This proposition is open to dispute.For example, action is often taken to avoid the possibility of harm — by regulation or prohibition —even though the likelihood of injury is remote, because of the seriousness of the potential outcome. Inmedical ethics, this is the principle of first, do no harm (primum non nocere). Two examples follow:(i) the antibiotic chloramphenicol was known to be highly effective in the treatment of variousinfections, including typhoid fever, but on rare occasions it induced bone marrow aplasia; despite thelow likelihood of this side-effect — about 1 in 250,000 — the use of chloramphenicol was restrictedto only a few life-threatening infections (e.g. typhoid fever), and it is now almost never used becausesafer effective alternatives are available; (ii) over recent years, there has been a flood of publicityover global warming and greenhouse gas emissions. A causal or direct relationship betweengreenhouse gases (such as CO2 and methane) and climate change is open to argument, and Earthundergoes repeated cycles of natural cooling and warming; in this respect, there is also evidence thatmelting of the Antarctic ice cap has been going on for some thousands of years, and global warmingfor over 100 years. Nonetheless, the consequences of inaction over greenhouse gas emissions arepotentially so serious that strategies to reduce the release of these gases into the atmosphere areentirely appropriate, despite uncertainty over the link between them and global warming.

(f) General observations on induction of mesothelioma by asbestos, especially the amphibolevarieties of asbestos such as crocidolite and amosite

(i) The linkage between the amphibole varieties of asbestos, and commercial chrysotile, and thesubsequent development of malignant mesothelioma is well established and is not in dispute.

5.79 This link is generally accepted as causal; in this respect, asbestos fulfils all The Bradford HillCriteria for the establishment of causality [44].

(ii) There is a dose-response relationship between cumulative exposure to asbestos and thesubsequent incidence of mesothelioma in asbestos-exposed cohorts or populations; the incidence isalso related to time since exposure, so that early exposures are more significant for mesotheliomainduction than later exposures, other factors being equal.

5.80 This relationship is expressed by the Peto model and its various modifications:

I = K*F*(Tp – [T – D]p)

where I = incidence; K depends on fibre type, mix, size and other site-specific variables;F = intensity of exposure in f/ml; and D = years of exposure. For the purposes of modelling, T can bereplaced by (T – 10) to build in a minimum 10-year lag-time, and the cubic power of time (T3) is oftenused, so that:

I = K*F*([T – 10]3 – [T – 10 – D]3)

WT/DS135/RPage 260

An important aspect of this model is that early exposures are more significant for mesotheliomainduction than later equivalent doses.

5.81 Of the variables D, F and K, it is D that is the most accurately measurable, whereas the valuesfor K and F are often unknown, though some estimates of F can be made from the type of workactivity. When there are multiple periods of employment for which the type of work is similar foreach, one can assume that the value for each of F1, F2, F3 ... Fn remains constant, which also applies toK1, K2, K3 ... Kn, so that:

I ∝ ([T – 10]3 – [T – 10 – D]3)

In practice, a simpler equation can be used: I = ctk

where the constant c is dependent on exposure, usually taken as proportional to the intensity ofexposure multiplied by its duration (i.e. cumulative exposure), with weightings for different fibretypes; the power k remains about 3.5, or 3 for short periods of exposure. As de Klerk and Armstrong[135] state:

"The model predicts that risk is increased after each increment of exposure by an amount proportionalto the level of exposure and the cube of time after that. In terms of the multistage model of cancer, itimplies that asbestos acts at the first stage of a 4-stage process. ... The model predicts that incidence ismuch more dependent on early or low levels of exposure and increases less rapidly as exposurecontinues to increase, depending mainly on time since first exposed." [p 232].

5.82 When one is faced with multiple exposures to asbestos, the following points emerge,specifically for mesothelioma induction, provided that the characteristics and time for each exposureare appropriate for a biological effect: (i) it is not valid to point to one exposure among the others andincriminate it as the sole cause of a mesothelioma, with exoneration of the other exposures; (ii) it isnot valid to point to one exposure among the others and exonerate it from a causative role in thedevelopment of a mesothelioma, and to incriminate all the others; (iii) when there are multipleepisodes of exposure as a background to a mesothelioma, it is often the case that each exposure inisolation would be sufficient for attribution of the mesothelioma to asbestos, with the provisosmentioned above (characteristics and times of exposures). When each exposure among others isappropriate for mesothelioma induction if the particular exposure occurred alone, it is not logical tostate that this exposure — which could have a biological effect in isolation — has no effect when incombination. In such circumstances, it is not the presence or absence of an effect that is in question,but the magnitude of each effect in proportion to the others.

5.83 A dose-response relationship has been observed with both estimates of airborne exposure toasbestos [136], and quantitative and qualitative fibre burden analysis of the asbestos content in humanlung tissue of mesothelioma patients [3, 25, 137, 138]: e.g. see Rogers et al. [3], and, more recently,Williams et al. [138], who noted in 1997 that:

"It was shown there that while the relative risk of all three diseases [i.e. asbestosis, mesothelioma andlung cancer] increased with increasing exposure, the relative risk of malignant mesothelioma is greaterat low levels of exposure when compared with the risk of asbestosis but is lower at very high levels ofexposure." [p. 39].

5.84 In their study on the relationship between lung asbestos fibre type and the lung tissueconcentration of asbestos versus the relative risk of mesothelioma, Rogers et al. [3] made thefollowing comment:

"Fiber content in the lung depends on both the amount of fiber deposited and the amount cleared. Theamount deposited depends on duration and intensity of exposure in the occupational or general

WT/DS135/RPage 261

environment. Clearance rate is thought to be dependent on the amount deposited at any point in time,i.e, clearance is exponential. Thus, the same fiber content in the lung at death or time of resection maybe achieved from a high initial deposition, followed by absence of deposition and absence of clearanceover a long period of time, or by a continuous deposition at a lower level, with or without clearance.Since detailed mechanisms of mesothelioma initiation and progression are not known, 'dose' asestimated by final lung fiber content may not relate to the 'dose' required to produce mesothelioma. Itis thus possible that a high lung fiber content in a mesothelioma case may represent continuingaccumulation of fibers after a lower level of fibers had produced malignant change. It is more likely,however, that the malignant change did not occur until the fiber content reached a sufficiently highlevel." [p 1913].

(iii) The dose-response relationship between the amphiboles and mixtures of asbestos types islinear at high exposures [15]

5.85 For example, please see EHC 203 and Table 4.

(iv) This dose-response relationship between asbestos exposure and the risk of mesothelioma hasalso been detected at low levels of exposure, which overlap with environmental exposures.

TABLE 4: INCIDENCE OF MESOTHELIOMA IN OCCUPATIONALLY EXPOSED GROUPS BY FIBRE TYPE ANDTIME SINCE FIRST EMPLOYED

Fibre type Industry Years since firstemployed

Rate per millionperson–years

Mixed: crocidolite, amosite andchrysotile

Manufacture textilesand insulation

20-2425-3030+

152017103180

Mixed, mainly amosite Insulation workers 20-2425-2930-3435-3940-4445+

29015502760630063308110

Mixed: crocidolite andchrysotile

Fibrous cement manufacture 20-2425-2930-34

270063009600

Chrysotile, some crocidolite Textile manufacture 20-2425-2930-3435-3940+

108143

1156493

1774Amosite Insulation manufacture 20-24

25-2930-3435+

744262350781842

Mixed Dockyards 20-2425-2930-3435-4040-4445-49

120410220370

12401510

Crocidolite Mining and milling 20-2425-2930-3435-39

900220030007000

From de Klerk and Armstrong [135].

WT/DS135/RPage 262

5.86 A recent case-control study [136] from France on the dose-response relationship between lowlevels of asbestos exposure and the odds ratio (OR) for mesothelioma showed a clear dose-responserelationship between estimated cumulative asbestos exposure and the OR for pleural mesothelioma.In the final paragraph of the article, the authors stated:

"We found a clear dose-response relation between cumulative exposure to asbestos and pleuralmesothelioma in a population-based control study, with retrospective assessment of exposure. Asignificant excess of mesothelioma was observed for levels of cumulative exposure that were probably farbelow the limits adopted in many industrial countries during the 1980s." [last sentence of abstract].

Although some concerns have been expressed about this type of investigation [139], it is my opinionthat these points were addressed in the original paper [136], and they are common and intrinsic toepidemiological studies of this type — e.g. see Camus et al. [140, 141]. This study [136] found anOR for mesothelioma of 4.2 [95 per cent CI 2.0-8.8] at estimated cumulative exposures of 0.5-0.99fibre-year,17 with elevation of the OR at about 0.5 fibre-year.

5.87 In a fibre burden study on mesothelioma patients, Rödelsperger [137] observed that:

"A significantly increased OR [for mesothelioma] is obtained even within the very low concentrationrange of 0.1-0.2 F/µg [i.e. concentrations in the range of 100,000-200,000 fibres per gram dry lungtissue], which may be expected for about 5% of the population." [p lll] (which also corresponds to anestimated cumulative exposure in the range of about 1-2 fibre-years).

5.88 In a more recent study on mesothelioma cases (N = 66) and controls (N = 66), Rödelsperger etal. [25] found an OR for mesothelioma of 4.5 at fibre concentrations of 100,000 to < 200,000 pergram dry lung tissue (for fibres > 5 µm in length; 95 per cent CI 1.1-17.9). These authors alsorecorded an OR = 2.4 at concentrations of 50,000 to < 100,000 fibres per gram dry lung tissue (95 percent CI 0.8-7.6). The controls for this study — surgical lung resections mainly for lung cancer —would be expected to bias the OR towards 1.0 (i.e. underestimate the effect) [25], and hence the OR of2.4 probably represents a genuine doubling of risk or more at these low fibre concentrations.

"Even within the concentration interval of 0.1-0.2 f/µg dry weight [i.e. 100,000 to 200,000 fibres pergram dry weight], a significantly increased odds ratio of 4.5 was obtained. Previously, the samemethod of tissue analysis was used to estimate a 95 percentile of the amphibole fibre concentration of0.1 f/µg dry weight for persons without detectable exposure to asbestos at the workplace. Therefore,within the range of the normal background level [up to 300,000 fibres per gram dry lung in Germany],a positive dose response is observed." [p 191].

This study did not detect an increase in the OR for chrysotile or for other mineral fibres.

5.89 The risk detected by Rödelsperger et al. [25] appears to correlate reasonably well with theFrench case-control study reported by Iwatsubo et al. [136], which found an OR for mesotheliomaof 4.2 at estimated cumulative exposures of 0.5-0.99 fibre-year, with elevation of the OR atabout 0.5 fibre-year.

(v) No lower threshold (minimum) level of asbestos exposure has been delineated, below whichthere is demonstrably no increase in the risk of mesothelioma.

5.90 This observation is expressed by Hillerdal [20] in the following terms:

"There is no proof of a threshold value — that is, a minimal lower limit below which asbestos fibrescannot cause the tumour [i.e. mesothelioma] — and thus it is plausible that even such low exposure can

17Fibre-year = concentration of airborne asbestos fibres (f/ml) x years of exposure.

WT/DS135/RPage 263

cause mesothelioma (even if the risk is extremely low). Patients with mesothelioma whose lungs showfibre concentrations within the normal range cannot be dismissed as background cases, — that is, notdue to asbestos. ... The only way to prove such a hypothesis would be to compare the incidence ofmesothelioma in a group with such background exposure with the incidence in a truly non-exposedgroup. This is not possible, as no such group can be found." [i.e. the lung tissue of virtually allmammals contains some asbestos fibres derived from natural, environmental or occupational sources][p 510].

5.91 Points worth emphasis in Hillerdal's review [20] include reports of mesothelioma amongschool teachers (9/487 patients with mesothelioma in one reference), in jewellers, and in individualsexposed to asbestos insulation at home (6/262 patients with mesothelioma according to one reference).Hillerdal [20] also makes the point that low-level asbestos exposure "more often than not containspeak concentrations which can be very high for short periods" (e.g. airborne asbestos fibreconcentrations of up to 78 f/ml from sweeping asbestos from the floor [New Caledonia]).

(vi) The dose-response relationship for commercial Canadian chrysotile and mesotheliomaincidence is also linear at high levels of exposure.

5.92 For example, among the Quebec chrysotile miners and millers it has been noted that:

"All the observed 38 cases were pleural with the exception of one of low diagnostic probability, whichwas pleuro-peritoneal. None occurred in workers exposed for less than two years. There was a cleardose-response relationship, with crude rates of mesotheliomas (cases/thousand person-years) rangingfrom 0.15 with cumulative exposure < 3530 million particles per m3 (mpcm)-years (< 100 millionparticles per cubic foot (mpcf)-years) to 0.97 for those with exposures of more than 10 590 mpcm-years(> 300 mpcf-years)." [EHC 203, p 8].

(vii) So far as I am aware, there are no observational data on dose-response relationships betweenchrysotile only at low exposure levels and mesothelioma incidence; in this respect, estimates arebased on extrapolation of a linear dose-response line from high exposures down to low exposures.

"Overall, the available toxicological data provide clear evidence that chrysotile fibres can causefibrogenic and carcinogenic hazard to humans. The data, however, are not adequate for providingquantitative estimates of the risk to humans. This is because there are inadequate exposure-responsedata from inhalation studies, and there are uncertainties concerning the sensitivities of the animalstudies for predicting human risk." [EHC 203, p 7].

5.93 Because of the lack of such data, no definite threshold for chrysotile in relation tomesothelioma and lung cancer has been delineated: According to EHC 203 (p 144):

"(a) Exposure to chrysotile asbestos poses increased risks for asbestosis, lung cancer andmesothelioma in a dose-dependent manner. No threshold has been identified for carcinogenic risks."

5.94 In summary:

TABLE 5: ASBESTOS-RELATED DOSE-RESPONSE RELATIONSHIPS FOR MESOTHELIOMA

Amphiboles Chrysotile

Heavy exposure Dose-response effect; linear Dose-response effect; linearLow-level exposure Dose-response effect No dataThreshold No threshold delineated No threshold delineated

(viii) To the best of my knowledge, there are no observational data on the interactive effect of low(or for that matter, high) concentrations of inhaled chrysotile fibres only, when these are

WT/DS135/RPage 264

superimposed later and separately upon a pre-existing amphibole ± chrysotile burden within lungtissue (?superimpositional additive or multiplicative effect).

5.95 For example, it has been estimated that up to 15-20 per cent of men in industrialized societiesmay have sustained occupational exposure to asbestos (chrysotile/amphiboles). Rödelsperger et al.[137] indicate that fibre concentrations of 100,000-200,000 amphibole fibres per gram dry weight lungtissue may be expected for about 5 per cent of the population in Germany. We do not know what theeffect of subsequent chrysotile fibre inhalation on top of this type of amphibole burden might be.

"Data were analysed on a case-referent basis, to relate relative risks of mesothelioma to dose of fibre, asmeasured both by lung content and estimated airborne exposure. Multivariate analysis of cases found adose response relationship for lung fibre content of crocidolite, amosite and chrysotile and thedevelopment of mesothelioma. Either a multiplicative or additive model could be used to fit therelative risk/dose coefficients for the various asbestos types. A progressive increase in relative riskwith increasing fibre content was reported for all fibres ... . Tests for trend were highly significant in allcases." [NICNAS 99, p 61].

(ix) There is a long lag-time between asbestos exposure and the subsequent diagnosis ofmesothelioma (10 years as a minimum; usually in the range of 20-40 years). It follows that themesotheliomas encountered in the 1990s and the incidence of mesothelioma in various nations are aconsequence of exposures, especially occupational exposures, sustained from the 1940s through to the1970s and even beyond.

5.96 Exposures from the 1940s through to the 1980s usually involved one or more of theamphibole varieties of asbestos. For example, asbestos-cement building products used in Australiausually contained one or more of the amphibole varieties of asbestos, namely crocidolite or amosite,or both, at different times; in this respect, the use of crocidolite in the products was discontinuedin 1966, and amosite in 1984.

5.97 Peto et al. [24] make this point in the following terms:

"The extraordinarily high mesothelioma incidence throughout Western Europe in men bornaround 1945-50 reflects the extent of asbestos use in the 1960s and 1970s at the beginning of theirworking lives. Annual raw asbestos imports to European Union countries peaked in the early tomid 1970s and remained above 800 000 tonnes per year until 1980, falling to about 100 000 tonnesby 1993 (European Commission, 1996). Increasingly stringent exposure limits were enforced in themanufacture of asbestos-containing products over this period, but exposure to users of such materials,particularly in the building industry, remained virtually uncontrolled in many countries. Chrysotileasbestos products are still widely used in several European countries, and maintenance or demolitionwork on older buildings may result in substantial exposure to amphiboles as well as to chrysotile. Wehave not included men born after 1955 in our projections, but the effects of asbestos exposure duringthe 1980s and 1990s, although not yet apparent, may prove considerable." [p 670].

(x) Properties of asbestos fibres that determine carcinogenicity

5.98 As indicated in the documents submitted to the WTO Panel, the properties of asbestos fibresimplicated for mesothelioma induction (and, possibly, lung cancer and other disorders), can besummarized as the three Ds:

5.99 Dose: this issue is covered in the preceding sections [(f)(ii)to (vi)].

5.100 Dimensions: according to The Stanton Hypothesis, the carcinogenicity of asbestos fibresappears to reside primarily in long thin fibres (length > 5 µm and especially > 8 µm, and in the rangeof 10-20 µm, and diameters < 0.25 µm) — e.g. see Pott [142]. On the other hand, shorter fibresappear to be less carcinogenic, although data indicate that tremolite fibres > 4 µm in length

WT/DS135/RPage 265

and < 1.5 µm in diameter produce malignant mesenchymal tumours when implanted into the pleuralcavities of rats [2]. On the other hand, very short-length fibres appear to have little carcinogenicactivity, although Churg [143] comments on fibre dimensions in the following terms:

"There has been extensive investigation of the relation of mesothelioma induction and fiber size inexperimental models. Using intrapleural inoculation of different types of fibers with different sizedistributions, Stanton et al. concluded that long, thin (i.e., high aspect ratio) fibres were much morepowerful mesothelial carcinogens than were short, thick fibers and that fiber type was less important.The exact size of fiber that qualifies as long and thin is unclear: fibers ... longer than 8 µm and widthsnarrower than 1.5 µm are usually cited from Stanton's work, but the same experiments show that fiberswith lengths greater 4 µm and widths less than 0.25 µm were also effective carcinogens. The Stantonhypothesis has been supported by animal inhalation experiments using size-separated fibers: fewmesotheliomas were found with either amosite or chrysotile prepared to contain few fibers longerthan 5 µm.

Human data on the question of fiber length and mesothelioma are equivocal. The tremolite found as anatural constituent of chrysotile ore is a relatively short, thick fiber compared with commercial amositeor crocidolite, and if one attributes 'chrysotile-induced' mesotheliomas in man to the tremolitecomponent, the differences in mesothelioma do correlate with fiber size. However, attempts to provethis proposition directly have produced equivocal results ... McDonald et al. concluded that thenumber of fibers longer than 8 µm explained most mesotheliomas and that chrysotile played no role.However, Rogers et al. found that fibers both longer and shorter than 10 µm, including chrysotilefibers, played a role, although long fibers were generally more important. The problem with both ofthese studies is that most patients with mesothelioma have had occupational asbestos exposure, andfibers in lungs from those with occupational exposure are always longer than fibers in the generalpopulation; thus the same result would have been obtained if the test group were exposed but had nodisease or some disease other than mesothelioma. My colleagues and I have attempted to circumventthis problem by comparing fiber sizes in a chrysotile mining and milling cohort and a cohort with heavyamosite exposure, using exposed workers with no disease as the control group. In neither cohort couldwe show that fibers in mesothelioma cases were significantly longer and thinner than those in the otherdisease categories or even in the disease-free workers." [p 353].

5.101 In other words, it is possibly the bio-persistence of amphibole fibres that is important formesothelioma induction, rather than precise fibre dimensions.

5.102 Durability (bio-persistence): the greater mesotheliomagenic (mesothelioma-producing)potency of the amphiboles in comparison to chrysotile is widely ascribed to greater persistence of theamphiboles in tissues, with significantly longer half-lives than chrysotile (please see later discussion,Section (g)(v)). On the other hand, it is conceivable that the same effect might be achieved bysustained inhalation of chrysotile over a prolonged time interval or, possibly, shorter, but more intenseexposures so that the chrysotile fibres persist despite shorter half-lives than the amphiboles.

(xi) There is general though not universal agreement of a differential potency between theamphiboles versus chrysotile for mesothelioma induction

5.103 In this respect, the amphiboles are substantially more potent, with estimates ranging from2-4X, to 10X, to 12X on a fibre-for-fibre basis, to 30X, to a 30-60X greater potency, or more(e.g. please see EHC 203). A minority view that the amphiboles in chrysotile have roughly equalmesotheliomagenicity is not supported by the prevailing evidence for humans. Althoughacknowledging the greater potency of the amphiboles for mesothelioma induction, some argue thatchrysotile is of equal or greater importance overall, because chrysotile accounts for > 95 per cent ofworld asbestos production. According to this perspective, commercial chrysotile is a weakercarcinogen on a fibre-for-fibre basis, but this lesser potency is multiplied across a much greatertonnage, leading to an overall equivalent or greater effect [144].

WT/DS135/RPage 266

(xii) Tobacco smoke plays no role in the development of mesothelioma at any anatomical site —unlike the synergy between asbestos and tobacco smoke for the causation of asbestos-related lungcancer (see section (i)(i) below).

(g) Commercial Chrysotile and Mesothelioma Induction

(i) There is general agreement that commercial chrysotile has the capacity to inducemesothelioma in experimental animals and humans

5.104 There is dispute, however, over which fibres in commercial chrysotile are implicated (i.e. thepredominant chrysotile or the trace quantities of fibrous tremolite).

(ii) Canadian chrysotile contains trace amounts of tremolite, including fibrous tremolite, as acontaminant [2, 10, 13, 14, 145-148]

5.105 The amount of tremolite appears to vary from one sample to another, but is generally < 1 percent (please see EHC 203).

(iii) It has been argued that the occurrence of mesotheliomas among the Quebec chrysotile minersand millers is a consequence — not of the chrysotile per se — but of the coexistent trace quantities oftremolite (a non-commercial amphibole).

5.106 Analysis of the asbestos fibre content of lung tissue from this cohort demonstratesdisproportionately high concentrations of tremolite in comparison to chrysotile; this appears torepresent a bio-accumulation phenomenon whereby chrysotile is cleared from lung tissue more rapidlythan the tremolite, so that the tremolite not only persists but increases in proportional concentration.In this respect, the tremolite content of the lung tissue can be used as an index on the past chrysotileexposure and some claim that the incidence of mesotheliomas in the same cohort can be relateddirectly to the tremolite content [13, 14].

(iv) It is known that fibrous tremolite has the capacity for mesothelioma induction

5.107 Mesotheliomas related to the use of tremolite in whitewash or stucco have been reported inTurkey, Greece, Cyprus and Corsica [149-152] (for additional references, see Hillerdal [20]).

"Tremolite asbestos, a minor component mineral of commercial chrysotile, has also been shown to becarcinogenic and fibrogenic in a single inhalation experiment and an intraperitoneal injection study inrats. Exposure/dose-response data are not available to allow direct comparison of the cancer potency oftremolite and chrysotile." [EHC 203, p 6].

5.108 Tremolite has also been implicated in lung cancer and mesothelioma induction in a group ofvermiculite miners in Montana [2, 16, 153, 154]. It appears that these miners were exposed only totremolite-actinolite fibres. The group was shown to have a:

" ... very high lung cancer incidence (standard mortality ratio [SMR] 285 ...), as well as four cases ofmesothelioma and eight of pneumoconiosis. Examination of sputum samples from all but three(170/173) current workers demonstrated asbestos bodies (AB) in 75%, the numbers showing a closeparallel with cumulative exposures in fibre-years." [2] [p 493].

5.109 Case [2] has extensively reviewed the biohazards of tremolite, including epidemiologicalinvestigations in humans and experimental data on animal models. In his review, he emphasized thepathogenicity of the tremolite found in Quebec chrysotile samples, especially at Asbestos and in theThetford mine:

WT/DS135/RPage 267

"Tremolite was not identified in Montreal air, was just detectable (0.2 fibres/l) in Asbestos, and wasone order of magnitude higher in Thetford mines (still only 1.5 fibres/l or 0.0015 fibres/cc ...)." [pp496-497].

5.110 He also favoured the expression "chrysotile/tremolite" for Quebec chrysotile:

"As to the separate issue of 'chrysotile vs. tremolite', few would dispute the abilities of both to producelung cancer and asbestosis, again in sufficient exposure dose. The weight of epidemiological, animal,and, especially, lung internal-dose biomarker studies leads to the inevitable conclusion that it is thetremolite 'component' of Quebec chrysotile which causes mesothelioma [but please see later discussionin this report]. It is unfortunate that adequate terminology for tremolite-contaminated chrysotile has notbeen introduced: I for one would favour the simple compound phrase 'chrysotile/tremolite'." [p 500].

5.111 Case [2] also states:

" ... it becomes important to know to what degree 'chrysotile-in-place' is really 'chrysotile/tremolite-in-place'. No easy answer can be expected: both bulk analyses and air sampling, even with analyticalelectron microscopy, can miss very low levels of tremolite. Studies in the Quebec mining districtindicate that, at the very least, such low levels (roughly 0.0015 fibres/cc) can induce biological effects(i.e., pleural plaques). Unfortunately, only expensive in vivo animal bioaccumulation assay systemscan truly answer the question: the alternative is to wait 40 to 50 years for the next wave of asbestosdisease — which is likely to occur mainly among present-day asbestos abatement workers and to somedegree in custodial personnel and other tradesmen" …. [p. 500].

(v) Clearance of chrysotile from lung tissue

5.112 It is well known that chrysotile fibres are cleared more rapidly than amphiboles, especially inlong-term studies [145]. Clearance of amphibole fibres does occur and the clearance mechanismsappear to be more effective for short fibres (for both chrysotile and the amphiboles) so that the meanlength of retained fibres increases over time. Churg and Vedal [155] calculated a half-life in lungtissue of about 20 years for amosite. Estimates of the tissue half-life for crocidolite fibres have beensomewhat shorter (in the order of 5-10 years) [156-158], and de Klerk et al. [158] could find nodifference between the clearance rates for long and short fibres. Oberdörster [159] estimates humanclearance half-times to be about 90-110 days for chrysotile and 200-1500 days for crocidolitefibres > 16 µm in length, based on extrapolated rat and primate inhalation data.

5.113 It has been claimed that chrysotile is cleared from lung tissue within 28-48 hours ofinhalation. This claim seems extraordinary and begs the question: why, if chrysotile is cleared fromlung tissue so rapidly, is it still demonstrable in human lung tissue many years or decades aftercessation of inhalation of commercial chrysotile (or mixtures of asbestos types)? For example, in oneof my recent referral cases — an elderly man with lung cancer who sustained exposure from mixingloose asbestos and sweeping up dried insulation materials — an asbestos fibre analysis carried out onlung tissue resected 16 years after his exposure stopped showed a total asbestos fibre count of8,440,000 fibres/gram dry lung (> 1 µm in length; aspect ratio ≥ 3:1), made up by 6,250,000 chrysotilefibres + 940,000 tremolite fibres + 940,000 anthophyllite fibres + 310,000 crocidolite fibres (the 24year lag-time is enough for a carcinogenic effect).

(vi) The Quebec chrysotile cohort

4.114 In an analysis of mesotheliomas among the Quebec chrysotile miners and millers, up to 1997,McDonald et al. [13, 14] reported 38 mesotheliomas, and most of these occurred after prolonged andheavy exposure, especially at the mine where the greatest concentrations of trace tremolite occurred(Thetford). For example, these authors [13] recorded the breakdown of the mesotheliomas shown inTable 6 (below).

WT/DS135/RPage 268

5.115 McDonald et al. [13] identify two main reasons for the low mesothelioma rate from the fivesmallest mines (1 case only among 6010 person-years, equivalent to 166 cases per million person-years): firstly, workers within this sub-group were younger than the remainder of the cohort;secondly, these mines had been opened recently so that "there were inadequate periods of latency". Asingle additional mesothelioma shortly after completion of the study would erase the difference inincidence rates between the five smallest mines and the main complex. McDonald et al. [13] go on toindicate that the other rates are "reasonably comparable". In comparison to the Thetford maincomplex, there were relatively few mesotheliomas among workers at the asbestos mine and mill(23 versus 8), despite nearly equivalent person-years of observation; in addition, asbestos fibreanalysis on lung tissue demonstrated crocidolite and amosite in five out of the eight cases from themine and mill at Asbestos and in two out of the five mesotheliomas from the Asbestos factory(Table 7, below). In focussing on the Thetford mines group, it was noted that most of themesotheliomas came from the five central mines (Area A; Group C) as opposed to the 10 peripheralmines (Area B; Group P), so that the odds ratio for mesothelioma for Group C plus employees whohad jobs in both Area A and Area B (Group M) was 2.50 (based on net service; 20 adjusted years), incomparison to an odds ratio of 0.80 for Group P.

TABLE 6: MESOTHELIOMAS AMONG QUEBEC CHRYSOTILE MINERS AND MILLERS, 1997

Number ofmesothelioma deaths

Subject-years(000s)

Rate(per 100,000 subject-years)

23 65.14 35.3Thetford Mines:

Main complex and theoldest of the smaller mines

The five smallest mines 1 6.01 26.6

8 60.64 13.2Asbestos:

Mine and millFactory 5 10.84 46.2

From McDonald et al. [13].

5.116 The clear implication of this complex and sophisticated study is that the risk of mesotheliomawas related strongly to years of service in the central area at Thetford where geological factors "inArea A would probably result in tremolite, some in fibrous form, being mined with the ore". Inaddition, the mesothelioma rate for miners and millers was > 2.5 times higher at Thetford mines(excluding the smallest mines) than at Asbestos, and this difference was also attributed to differencesin the amount of fibrous tremolite in the ores. Despite these differences within the cohort for thedistribution of mesothelioma related to chrysotile and tremolite (and also to crocidolite and amosite atthe Asbestos factory and the Asbestos mine and mill), the results indicate that Quebec chrysotile — onaverage contaminated by fibrous tremolite in small amounts — is capable of mesothelioma induction:the Abstract describes 25 mesotheliomas from the Thetford mines, representing a mesothelioma rateof 337 per million person-years, which is substantially (almost 20X) higher than the mesotheliomaincidence rate of about 17 per million per person-years for men in British Colombia and the USA in1982 and 1973-1984 respectively, and well above the background rate for spontaneous mesotheliomasof 1-2 per million person-years.

WT/DS135/RPage 269

TABLE 7: ASBESTOS FIBRE CONCENTRATIONS IN LUNGS AT AUTOPSY FROM 21 MESOTHELIOMACASES AMONG QUEBEC CHRYSOTILE MINERS AND MILLERS

(FIBRES PER µg: GEOMETRIC MEANS)

Place of employment No. of cases Chrysotile Tremolite Crocidolite Amosite

Mines and mills

Thetford Mines 14 12.8 104.1 0 0

Asbestos 5 4.3 7.5 1.7 0.3

Factory

Asbestos 2 2.1 0.5 6.4 0.3

Table from McDonald et al. (1997): Table 2 in the original reference. See also Table 1 in the original.In calculating geometric means, a zero count has been replaced by half the detectable limit.For crocidolite and amosite, all counts were zero: i.e. below the detectable limit.For fibre counts/g lung tissue, multiply the figures by 106.

5.117 In the final two paragraphs of the paper, McDonald et al. [13] comment as follows:

"The tremolite hypothesis, if correct, has several important implications. First, it supports the widelybut not universally held view that most, if not all, asbestos-related mesotheliomas are caused byamphibole fibres. This in turn points to fibre durability and biopersistence as critical factors inaetiology ... a point of even greater relevance in assessing the safety of man-made mineral fibres.Second, it implies that uncontaminated chrysotile carries very little risk of mesothelioma. In Asbestos,exposures were not to uncontaminated chrysotile, but also to some tremolite and crocidolite, yet amongthe miners and millers only five deaths from a total of over 3,300 can be confidently attributed to theirwork.

At present-day levels of dust control the mesothelioma risk must be vanishingly small. Even so, itremains desirable to minimise, perhaps by screening, the contamination of commercial chrysotile byamphibole fibres, however difficult this may be." [p 718].

5.118 Despite the importance of this study by McDonald et al. [13], the following comments canalso be made:

• The number of mesotheliomas in all groups except for the Thetford main complexwas small (1, 8 and 5 mesotheliomas respectively; please see Table 6 above). In thisrespect, misdiagnosis or misclassification of the mesotheliomas according to theplaces worked could significantly affect the results, although there is no evidence thatthis happened; however, the probability for the diagnosis of mesothelioma also varied,with a high probability in 19 cases, moderate probability in 14, and a low probability(though considered more likely than not) in five; of these 38 cases, only 18 had beencoded on the death certificate to ICD 163, and the rest to a variety of other diagnosticcodes. Furthermore, in analysing the mesotheliomas according to Area A versusArea B at the Thetford mines (groups C, M and P), the numbers were 104 for groupC, 69 for group P and 35 for group M; McDonald et al. noted that the odds ratio forgroup P was unstable as shown by the "very wide confidence intervals, and as thepoint estimate is well below unity it is quite unrealistic ".

• The low incidence of mesotheliomas in the Quebec chrysotile cohort appears toparallel similar low incidence rates for asbestosis and lung cancer for the same cohort

WT/DS135/RPage 270

[160, 161]; the incidence rates for lung cancer and mesothelioma appear to bedifferent in other chrysotile-exposed cohorts.

5.119 For these reasons and because of the different rates of various asbestos diseases (asbestosis,lung cancer and mesothelioma) between the Quebec cohort and other groups of workers, I would bereluctant to recommend national policies from the findings in this cohort in isolation, and I would lookfor coherence of the evidence across different cohorts and studies.

5.120 In relation to the Quebec cohort, there is an important error in the Canadian reply toQuestion 4 (see Annex II) from the European Communities, where the following statement is made:

"Regarding asbestos-related mesothelioma, a number of studies have demonstrated cogently that thistype of cancer is almost exclusively linked to exposure to amphiboles. Cases of mesothelioma inchrysotile asbestos miners in Quebec are quite rare — in a cohort of 11,000 workers who were verycarefully tracked (in the McDonald study), there were no more than 50 or so cases over severaldecades. Exhaustive research on their employment history revealed that most of the cases were relatedto short-term exposure to commercial amphiboles. For example, during World War II, some of theminers with mesothelioma had worked in plants manufacturing products for the Allied Forces andamphiboles imported into Canada had been used to make a variety of products, including gas masks, toassist in the War effort."

5.121 The statement that "most of the cases were related to short-term exposure to commercialamphiboles" is incorrect and misleading. As demonstrated in the study by McDonald et al. [13], mostof the mesotheliomas occurred among chrysotile miners who worked at the Thetford main complex,without exposure to commercial amphiboles such as crocidolite or amosite. This is clearly shown inTable 7 (above), slightly modified from the paper by McDonald et al. [13] where fibre burden analysison lung tissue from 14 mesothelioma cases from the Thetford mines showed both chrysotile and ahigh concentration of tremolite, with a zero count for the commercial amphiboles crocidolite andamosite. The point to be emphasized is that the mesotheliomas from the Thetford mines were notrelated to commercial amphiboles such as crocidolite or amosite, but to chrysotile with its content offibrous tremolite.

(vii) As discussed earlier, a dose-response relationship between the incidence of mesothelioma andcumulative asbestos exposure has been demonstrated for commercial chrysotile.

5.122 Mesotheliomas have also been produced in experimental animals by implantation andinhalation of chrysotile (presumably also containing trace amounts of tremolite). Mesotheliomas canalso be induced in rats by intraperitoneal injection of chrysotile, with evidence of a dose-responseeffect [1] (see also bibliography for EHC 203).

"In non-inhalation experiments (intrapleural and intraperitoneal injection studies), dose-responserelationships for mesothelioma have been demonstrated for chrysotile fibres." [EHC 203, p 5].

(viii) Chrysotile is also known to be toxic to a variety of cell lines in vitro, with induction of avariety of chromosomal alterations (e.g. please see EHC 203, pp 69-102).

(h) Other Chrysotile-Exposed Cohorts and Studies

5.123 In addition to the Quebec chrysotile miners and millers, mesotheliomas have also beenreported among other workforces apparently exposed only to chrysotile, with no significant tremolite.

WT/DS135/RPage 271

(i) Russia

5.124 Chrysotile from the Urals region (Uralasbest) in Russia [162, 163] is said to represent purechrysotile. Although precise figures for the mesothelioma incidence in this area are difficult toprocure, Kogan [164] makes the following comment in a recently-published textbook on occupationallung diseases:

"In the Middle Ural mountains, the main asbestos mining region in Russia, only chrysotile asbestos isproduced. In the 50 districts of this region, the mortality from mesothelioma over a 10-year period wassix-fold higher than the average rate in the Sverdlovsk region, an area of negligible asbestos mining.Most with mesothelioma had worked at the asbestos mining and milling plants, or had lived in anadjacent town near old and very 'dusty' mills." ... . [p 251].

5.125 Because it is difficult to equate exposure levels in the Russian chrysotile industry with otherindustries (e.g. the airborne fibre concentrations at Uralasbest are usually expressed as gravimetricmeasurements), and I have been unable to ascertain the numbers of cases relative to exposure levels, Iconsider this evidence to be weak in comparison to other studies.

5.126 One might expect data on mesothelioma incidence in Central and Eastern European nations tobe of interest, from an assumption that some of these countries would have imported mainly chrysotilefrom Russia until the breakup of the Soviet Union. Unfortunately, it is difficult to evaluate nationalmesothelioma statistics, because a number of these nations also imported amphibole asbestos. Forexample, in Slovenia, the total consumption of asbestos (1947-1995) was 580,000 tonnes, of whichcrocidolite accounted for 37,133 tons, until its use was stopped in 1992 [165]. Similarly, the annualusage of asbestos in Bulgaria during the 1970s and 1980s reached approximately 32,000 tons ofchrysotile (mainly from Russia and Canada), together with about 1000 tons of crocidolite from Africaand 6000-7000 tons of Bulgarian amphibole material (anthophyllite and tremolite) [166]. In Poland,total consumption of asbestos for the manufacture of asbestos-cement products between the end of theSecond World War until 1993 was about 1.4 million metric tons which included about 8500 metrictons of amosite and approximately 86,000 metric tons of crocidolite [167].

(ii) Germany

5.127 The former German Democratic Republic (GDR): Sturm et al. [5, 7] have published data onasbestos-related diseases and asbestos types in the German State of Saxony-Anhalt. These authorspointed out that:

"All asbestos-based products were made from raw asbestos which was primarily imported from theformer Soviet Union, particularly from the Kiembay mining area in the Ural mountains (said torepresent pure chrysotile). Small quantities of long-fibred grades came from Canada (2,990 tonnesin 1989) and were mainly used for the manufacture of asbestos-cement pressure pipes free ofamphibole asbestos. This was a share of approximately 7% in total imports. We never obtained anyinformation about the Canadian mines from which the asbestos processed in the former GDRoriginated. ... However, several analyses carried out by the GDR Central Institute for IndustrialMedicine confirmed that both the Canadian and the Russian asbestos were pure chrysotile. In additionto these imports of chrysotile asbestos, smaller quantities of amphibole asbestos were imported. Forexample, in the period from 1980 to 1985, some 90 tonnes of anthophyllite were imported annuallyfrom Mozambique. This anthophyllite was used exclusively by a Berlin manufacturer and was of acid-proof products, similar to the way crocidolite had been used in previous years to produce filters, sealsand acid and lye-proof plastic materials. In Saxony-Anhalt, our region of work, these amphiboleimports did not have any significance from the point of view of industrial medicine" ... [p 318/173].

5.128 Between 1960 and 1990, a total of 1082 mesotheliomas was recorded in Saxony-Anhalt, andthese included 843 "proven asbestos-accepted mesotheliomas"; Table 8 from Sturm et al. [5, 7] gives

WT/DS135/RPage 272

a breakdown of 812 cases for which adequate data were available: 67 were said to follow exposure tochrysotile only, and 331 were associated with "chrysotile; possible amphiboles".

(iii) Italy

5.129 Two mesotheliomas have now been recorded among more than 900 workers employed at theBalangero mine and mill in Italy [168, 169]. EHC 203 gives the following summary:

"The cohort of chrysotile production workers employed at the Balangero mine and mill ... was almostexactly one tenth the size of the Quebec cohort. At the end of 1987, when 427 (45%) of the cohort haddied, there were two deaths from pleural mesothelioma, both in men employed for more than 20 yearswith cumulative exposure estimated respectively at 100-400 and > 400 f/ml years. One diagnosis wasconfirmed histopathologically, and one was based on radiological findings and examination of pleuralfluid. Fibrous tremolite was not detected in samples of chrysotile from this mine, but another fibroussilicate (balangeroite), the biological effects of which are not known, was identified in low proportionsby mass (0.2-0.5%). At a comparable stage in the evolution of the Quebec cohort, mesotheliomaaccounted for 10 of 4547 deaths, a lower but not dissimilar proportion." [p 112].

TABLE 8: MESOTHELIOMAS ACCORDING TO TYPES OF EXPOSURESTO ASBESTOS IN SAXONY-ANHALT

Amphiboles Amphibolesand chrysotile

Chrysotile;possible

amphiboles

Chrysotile Mean values

Age at beginning ofexposure

25 28 28 34 28

Duration of exposure 16 21 19 14 19Lethal period (years) 40 40 41 31 38Age of person dyingof mesothelioma

65 68 69 65 66

Number ofmesotheliomas

135 279 331 67 N = 812

Note: All types of application of asbestos with common addition of chrysotile fall under the heading "Chrysotile.Amphiboles possible" when previous admixture of amphiboles cannot be definitely excluded. From Sturm et al. [5, 7] .

(iv) China

5.130 At the XV International Scientific Meeting of the International Epidemiological Association(Florence, September 1999), Yano et al. [170] presented a paper on lung cancer incidence in a cohortof 515 male asbestos workers heavily exposed to chrysotile containing < 0.001 per cent tremolite, inChongqin; two mesotheliomas over 11,850 person-years of observation occurred in this cohort(discussion to the paper; assuming this rate to be representative, it would amountto 170 mesotheliomas per million person-years).

5.131 In a retrospective cohort mortality study of 1227 men employed at a chrysotile mine in HebeiProvince of China before 1972, Zou et al. found three deaths from mesothelioma (please seeEHC 203, p 120).

(v) United States

5.132 Two mesotheliomas have also been observed among the cohort of South Carolina chrysotiletextile workers — who used Canadian chrysotile — studied by Dement et al. [171, 172] (please seeEHC 203, p 115).

WT/DS135/RPage 273

(vi) Australia

5.133 There is also some indication of an increased frequency of mesothelioma among Australianbrake mechanics who were potentially exposed only to chrysotile from grinding of brake blocks thatcontained Canadian chrysotile (please see later discussion on friction products, and NICNAS 99 andAMR 99).

(vii) Zimbabwe

5.134 One pathologically confirmed case of mesothelioma has been recorded in association withoccupational exposures to asbestos in the Zimbabwe mines and/or mills, with one other case said toresemble mesothelioma radiologically (EHC 203, p 121).

(viii) Fibre burden studies on human lung tissue from mesothelioma patients

5.135 Fibre burden analyses also support the notion that some mesotheliomas occur in associationwith, or as a consequence of, inhalation of pure chrysotile.

5.136 Morinaga et al. [173] detected asbestos fibres in 19 of 23 mesothelioma studied; amphibolefibres were found in 13 cases, but six were found to have only chrysotile fibres (five pleuralmesotheliomas and one peritoneal mesothelioma). Nonetheless, the methodology for this study seemsunimpressive, with relatively small numbers of fibres analysed.

4.137 The 1991 paper by Rogers et al. [3] recorded a substantial number of mesothelioma patientsin whom the only detectable type of asbestos was chrysotile (Table 9), with evidence of adose-response effect as reflected in a trend to an increasing odds ratio (OR) at a relatively low fibreconcentration of ≤ 106 fibres per gram dry lung tissue (log10 = 5.5–6; OR = 8.67).

TABLE 9: DISTRIBUTION OF FIBRE CONCENTRATION: TRANSMISSION ELECTRONMICROSCOPIC ANALYSIS, CHRYSOTILE ONLY (ALL LENGTHS)

Mesothelioma cases Controls Odds ratioNo. Percent No. Percent 95% Cornfield

Clf/g 0-200.000

1248.0

2683.9

log10 (f/g) 5.3-5.51

4.02

6.5 1.08 (0-17.95)

5.5-67

28.03

9.7 8.67(1.77-48.14)

6-6.53

12.0

6.5-71

4.0

7-81

4.0 Χ2 1 = 9.80(P<0.0005)(trend)

From Rogers et al. [3]. Cl: confidence interval; f/g: fibres per gram of dried lung tissue.

5.138 Finally, fibre burden studies have demonstrated that both chrysotile fibres and amphibolefibres can translocate from lung parenchyma to reach the pleura; EHC 203 summarizes these findingsin the following way:

"In a study of asbestos fibres in the lung parenchyma and the parietal pleura of 29 asbestos workers,

WT/DS135/RPage 274

Sebastien et al. (1980) found that chrysotile fibres predominated in the pleura and that amphibole fibrescould not be detected. A similar result was reported by Dodson et al. (1990). Kohyama & Suzuki(1991) found short chrysotile fibres in pleural plaques and in mesothelial tumours. In contrast, Boutinet al. (1993) found 0.21 x 106 fibres per g of parietal pleura and 1.96 x 106 in samples of lungparenchyma. Fibre concentrations were higher in subjects with a history of asbestos exposure and mostof the fibres were amphiboles. Churg (1994) reported detection of chrysotile fibres in the subpleuralparenchyma in chrysotile miners and millers." [pp 64-65].

(ix) Other Observations

5.139 Nicholson and Raffn [8] analysed mesothelioma risk over 40 studies for which little or noexposure information was available, using the excess numbers of lung cancers as a measure ofexposure and comparing the ratios of mesotheliomas to excess lung cancers across these studies.They suggested that:

" ... the ratio of mesothelioma to excess lung cancer is the same for exposures to 100% chrysotile(presumably Canadian chrysotile), 97%+ chrysotile, 100% amosite, and mixtures of chrysotile, amositeand crocidolite, within statistical uncertainty. Only 100% crocidolite exposures appear to have agreater ratio, about two to four times that of predominantly chrysotile. This relatively small differencein the potential for crocidolite to produce mesotheliomas compared with other fibre exposure cannotexplain the high risk seen in chrysotile exposures accompanied by a very small crocidolite exposure.The data speak strongly that much of the mesothelioma risk in predominantly chrysotile exposures isfrom the chrysotile." [p 402].

5.140 In other words, these authors appear to argue, like Smith and Wright [144], Stayner et al. [11],and Landrigan et al. [21] that although chrysotile may be cleared more rapidly from lung tissue thantremolite — and that tremolite can be used as an indicator of past chrysotile — it may not be valid toascribe all the mesothelioma risk to the tremolite and to ignore the far more numerous chrysotilefibres. Nonetheless, I do not find Nicholson and Raffn's argument to be persuasive, taking intoaccount the K values for different industries.

5.141 Therefore, it is my perception that epidemiological and experimental evidence clearlydemonstrates that Canadian chrysotile with its trace amounts of fibrous tremolite has the capacity formesothelioma induction. Although the tremolite may have a disproportionately large effect, it is myperception that the evidence does not allow one to conclude that the chrysotile has no effect onmesothelioma induction: there is evidence from other cohorts and studies that chrysotile per se canalso induce mesothelioma, even when tremolite is undetectable, and in experimental models inanimals, chrysotile is as carcinogenic as, and more toxic than, the amphiboles. However, there is alsogeneral agreement that in humans, chrysotile is substantially less carcinogenic for the mesotheliumthan the amphiboles, and my estimate is that it has a potency 1/10th – 1/30th the carcinogenicity ofcrocidolite, with amosite being less mesotheliomagenic than crocidolite but more carcinogenic thanchrysotile on a fibre-for-fibre basis. Amosite is an important factor in the incidence of mesotheliomain the United States, because of its widespread use in insulation materials from the 1960s [155, 174-176].

(i) Abestos and Lung Cancer

5.142 Still the focus of some controversy, this subject has been reviewed by Henderson et al.:(i) Henderson DW, Roggli VL, Shilkin KB et al., Is Asbestosis an Obligate Precursor for Asbestos-Induced Lung Cancer? In: Peters GA, Peters BJ, eds. Sourcebook on Asbestos Diseases, vol 11.Charlottesville: Michie; 1995;11:97-168 [177]; (ii) Henderson DW, de Klerk NH, Hammar SP, etal., Asbestos and Lung Cancer: Is it Attributable to Asbestosis, or to Asbestos Fiber Burden? In:Corrin B, ed. Pathology of Lung Tumors, New York: Churchill Livingstone; 1997:83-118 [131];(iii) Leigh J, Berry G, de Klerk NH, Henderson DW., Asbestos-Related Lung Cancer: Apportionment

WT/DS135/RPage 275

of Causation and Damages to Asbestos and Tobacco Smoke, In: Peters GA, Peters BJ, eds.Sourcebook on Asbestos Diseases, vol 13, Charlottesville: Michie; 1996:141-66 [178]; (iv) Multipleauthors. Consensus Report: Asbestos, Asbestosis, and Cancer: the Helsinki Criteria for Diagnosisand Attribution, Scand. J. Work Environ. Health 1997;23:311-6 [113].

5.143 Some salient features of asbestos-associated lung cancer include the following:

(i) Synergy between asbestos and tobacco smoke

5.144 Historically, most asbestos workers have also been cigarette smokers, and the lung cancer ratein virtually all cohorts is an outcome of the combined and synergistic effects of tobacco smoke andasbestos. Vainio and Boffetta [179] emphasize that asbestos and tobacco smoke are complexcarcinogens that can affect multiple steps in the multistage process of cancer evolution, and that thecombined effects will depend on the relative magnitude of each carcinogen at each stage; theinteractive effect ranges from less than additive to supramultiplicative, but the model for insulationworkers approximates a multiplicative effect (reviewed in Henderson et al. [131]). If the multistagemodel of carcinogenesis holds, and asbestos and smoking act at different stages, then a multiplicativerelationship follows [180]. Leigh et al. [178] have reviewed various models for the apportionment offractional contributions from cigarette smoke and asbestos towards the development of lung cancer.

(ii) Lung cancer incidence rates for asbestos-associated lung cancer vary greatly from one cohortto another

5.145 Please see following discussion.

(iii) Asbestos fibre type and lung cancer risk

5.146 The greater carcinogenicity of the amphiboles for the mesothelium in comparison tochrysotile appears not to extend to the induction of lung cancer [11]. In this respect, chrysotile isimplicated in one of the lowest rates of asbestos-associated lung cancer (in Quebec chrysotile minersand millers), but also the highest rate (in South Carolina asbestos textile workers who used Canadianchrysotile) [171]. The reasons underlying this ≥ 30-fold difference in lung cancer risk remainunknown (reviewed recently by McDonald [161]; please see also EHC 203). The risk of lung cancerin other asbestos-exposed cohorts is intermediate between these two extremes [15].

(iv) Dose-response relationship

5.147 In most studies, there is a direct and linear relationship between the relative risk of lungcancer and cumulative exposure to asbestos, including chrysotile and the amphiboles.

5.148 Accordingly, EHC 203 gives the following account:

"The slopes of the relationship between cumulative exposure to chrysotile and the relative risk of lungcancer are summarized in Table 23 for those studies that reported this information. These studies allexpressed this relationship using the following linear relative risk (RR) model:

RR = 1 + B x E

where B is the slope and E is the cumulative exposure to chrysotile asbestos expressed in f/ml-years.

The slopes from the studies of the mining and milling industries (0.0006 to 0.0017), the latter havingbeen estimated on a subset of the cohort on which the former was based, and the friction productionindustries (0.0005 to 0.0006) are reasonably similar. Hughes et al. (1987) in a study of cement workers(section 7.1.2.1b) reported a similar slope (0.0003) in one plant (plant 1) that only used chrysotile, and

WT/DS135/RPage 276

a nearly 20-fold higher slope (0.007) among workers only exposed to chrysotile in another plant(plant 2).

The slopes of 0.01 and 0.03 reported for the two studies of the chrysotile-exposed textile workersconducted on overlapping populations, as well as the slope of 0.007 from one of the two plants (plant 2)of cement workers in the study of Hughes et al. (1987), were an order of magnitude greater than thosereported for the other cohorts. It should be noted that the two textile cohorts were identified from thesame textile facility, but were based on different cohort definitions. Hence, it is not surprising that theslopes from these two studies were similar. The slopes in the studies of chrysotile-exposed textileworkers are also remarkably similar to those reported in other studies of textile workers with mixedfibre exposures (Peto, 1980; McDonald et al., 1983b; Peto et al., 1985). This similarity in findingsprovides some support for the validity of the slopes reported in the chrysotile-exposed textile cohorts.

The reason for the much higher slopes observed in studies of textile workers is unknown, althoughseveral possible explanations have been suggested. The first is that these differences might beattributed to errors in the classification of exposures in these studies. Particular concern has been raisedabout errors in the exposure assessment related to conversions from mpcm (mpcf) to fibres/ml that wereperformed, particularly in the mining and milling studies (Peto, 1989). Sebastien et al. (1989)conducted a lung burden study specifically designed to examine whether the differences in lung cancerslopes observed in the Charleston chrysotile textile cohort and the Quebec mining industries could beexplained by differences in errors in exposure estimates. Lung fibre concentrations were measured in:(a) 32 paired subjects that were matched on duration of exposure and time since last exposure; and(b) 136 subjects stratified on the same time variables. Both analyses indicated that theQuebec/Charleston ratios of chrysotile fibres in the lungs were even higher than the correspondingratios of estimated exposures. This finding was interpreted by the author as being clearly inconsistentwith the hypothesis that exposure misclassification could explain the large discrepancy in the lungexposure-response relationships observed in the two cohorts." [pp 118-119].

5.149 Boffetta [15] expresses the relationship in the following terms:

"A large number of studies have been conducted on lung cancer risk following asbestos exposure. Theinterpretation of their results is complicated by several factors: (i) dose, geological type of fibres andindustry are all important determinants of risk and are strictly correlated; (ii) the biologically relevantexposures occur 20 or more years before appearance of the disease, and their quantitative assessment isimprecise; and (iii) the role of potential confounders, in particular, tobacco smoking, can hardly beevaluated. In general, the risk of lung cancer is smaller in studies of miners and friction productmanufacturers, is intermediate in studies of asbestos-cement and asbestos product manufacturers, and ishighest in studies of asbestos textile workers. This likely reflects a stronger carcinogenic effect ofindividual, long and thin fibres, like those occurring in the textile industry, as compared to grouped,short and coarse fibres, like those occurring in mining.

Several cohort studies provide sufficient details to allow a quantitative evaluation of the risk of lungcancer from cumulative asbestos exposure. In all cohorts, the empirical relationship fits well a linearcorrelation with no threshold, which can be expressed as:

RR1 = 1 + K1*CE,

where RR1 is the relative risk of lung, CE represents cumulative asbestos exposure, expressed as fb/ml-yrs, and K1 is the industry-specific slope of the relationship (RR for the increase in 1 fb/ml-year ofexposure) for lung cancer and varies across cohorts. Similarly, the risk difference (RD1) can beexpressed as

RD1 = K1*CE*Exp,

where Exp is the number of expected cases of lung cancer. In other words, the number of cases of (ordeaths from) lung cancer attributable to asbestos exposure depends on the number of expected cases(deaths), the cumulative exposure, and the intrinsic carcinogenic potential of the exposure

WT/DS135/RPage 277

circumstance. The value K1 varies from 0.05-0.01 in cohorts of insulation and asbestos textile workersto 0.001-0.0005 in friction manufacturers and miners, while cohorts with mixed exposure have, in mostcases, intermediate values. ... While all estimated values of K1 are positive, the type of asbestos doesnot seem to be correlated to lung cancer risk.

In the interpretation of these results, however, one should consider several limitations. Most studies arebased on a small number of cases or deaths: for example, the risk estimate of 100 fb/ml-yrs for thecohort of asbestos textile workers presented by McDonald and colleagues (RR 2.4) has a 95%confidence interval from 1.7 to 3.8. Another source of uncertainty, and possibly bias, relates to theestimate of cumulative exposure: in the same cohort of asbestos textile workers, the range of RRsbased on the extremes of the distribution of possible exposure values is 1.3-6.7. For these reasons,several governmental and scientific committees have suggested to adopt an 'average' value of K1,independent from fibre type and circumstance of exposure ...: the most widely accepted value is 0.01which corresponds to an increase of 1% of the risk of lung cancer for each fb/ml-yr of exposure. ...

Tobacco smoking is the main cause of lung cancer, and this applies also to the cohorts of asbestos-exposed workers. Despite the limitations of the available studies, which limit the precision of theestimates of the combined effect of the two carcinogens, the risk from tobacco smoking seems to actsynergistically with that of asbestos exposure, according to a multiplicative model. ... The availabledata are consistent with the most widely accepted model of quantitative dose-response betweencumulative exposure to asbestos and lung cancer risk, which assumes a linear relationship with nothreshold. Alternative models, however, would also be consistent with the data. In particular, as noprecise data are available for cumulative exposures below 1 fb/ml, a model with a threshold at lowexposure cannot be rejected." [pp 473-475].

(v) Histological types of lung cancer

5.150 Although some studies have shown a relative excess of adenocarcinomas in proportion toother histological types of lung cancer, all of the major histological types occur among asbestosworkers in proportions equivalent to, or only slightly different from, those in the general population[112]. Therefore, the histological type of a lung cancer has no value in ascertaining whether or notasbestos has contributed significantly to the genesis of the cancer (reviewed by Henderson et al.[131]).

(vi) Lobar distribution and the central versus peripheral distribution of asbestos-related lungcancer

5.151 Some studies have reported a reversal of the upper lobe:lower lobe ratio for lung cancers inasbestos workers, in comparison to a reference non-exposed population. Recently, Lee et al. [181]addressed the lobar distribution of lung cancer in asbestos-exposed individuals and found that thetumours were predominantly located in the upper lobe (i.e. they did not find a reversal of the upperlobe to lower lobe ratio). The lobe of origin for a cancer has no value in ascertaining whether thecancer is likely to be asbestos-related. The distribution of lung cancer between the central versus theperipheral airways does not differ significantly in asbestos workers from a control non-exposedpopulation (please see Henderson et al. [131]).

(vii) Asbestos and lung cancer risk

5.152 Cumulative exposure versus fibrosis (asbestosis): as discussed already, most epidemiologicalstudies dealing with lung cancer risk in asbestos workers have reported a direct correlation betweenthe relative risk of lung cancer and cumulative asbestos exposure, although the slope of thedose-response line varies from one cohort to another. Most of the documents submitted to the WTOappear to agree on this relationship, the main area of uncertainty or dispute being the question ofwhether a threshold exists or not.

WT/DS135/RPage 278

5.153 However, Canada's answers to questions from the Panel and the European Communitiesappear to resurrect the fibrosis-cancer hypothesis, which postulates that asbestos does not induce lungcancer per se, but only through an obligate intermediary step of pulmonary fibrosis (asbestosis), sothat fibrosis becomes the determinator of lung cancer risk, not cumulative exposure:

"1. Canada does not disagree that chrysotile causes lung cancer. However, the way in whichexposure to chrysotile asbestos may increase the risk of lung cancer has not yet been fully explained; itcould be just an indirect cause. ...

2. The risk may become detectable in cases of long-term exposure to high levels, but it is by nomeans certain that chrysotile acts as a direct carcinogen or that it acts in the form of pulmonary fibrosis,which would be a precursor to neoplasia. In other words, exposure must be intense and long enough toinduce pulmonary fibrosis, which predisposes the pulmonary parenchyma to a high risk of cancer."

5.154 It is my perception that the fibrosis→cancer hypothesis represents a minority opinion: withsome prominent exceptions, most authorities in this area reject the fibrosis→cancer theory and focusinstead on the asbestos fibre burden in lung tissue as the main determinator for lung cancer risk, asdiscussed earlier in this report.

5.155 The fibrosis→cancer hypothesis is predicated upon three key but flawed studies:

• In the investigation reported by Kipen et al. [182], there was major problem with caseselection (only 138 cases out 450 — 31 per cent — had a tissue specimen withsufficient non-malignant tissue for assessment of fibrosis); in addition, thehistological criteria used for the diagnosis of asbestosis are unacceptable to mostpathologists — i.e. no asbestos bodies in some cases; fibrosis restricted to thesubpleural zone considered to be asbestosis — so that this study seems to havesuffered from an over-diagnosis of asbestosis [183, 184].

• As discussed in Section (e) above, the Hughes-Weill study [133] on chest X-rayopacities related to lung cancer mortality in New Orleans asbestos-cement workershad low statistical power, so that it had only a 40 per cent chance of detecting asignificant lung cancer standardized mortality ratio (SMR) of 1.5. Other studiesbased on X-rays have shown an increase in risk or mortality for lung cancer in theabsence of radiological asbestosis (e.g. Wilkinson et al. [185], Finkelstein [186] andde Klerk et al. [187]).

• The autopsy study on South African crocidolite miners reported by Sluis-Cremer andBezuidenhout [188] was also bedevilled by problems of selection (black peopleexcluded; autopsies on 36.7 per cent of deaths only; autopsies on cases for whichcompensation was sought). Analysis of the findings indicates that the effect ofduration of exposure (the most accurately measurable of the exposure variables) wasstill significant even after adjustment for the grade of asbestosis and other variables.This indicates that exposure to asbestos still had an independent effect on lung cancermortality even after adjustment for the grade of asbestosis, as in the study reported byWilkinson et al. [185]. In subsequent correspondence, Sluis-Cremer andBezuidenhout [189] conceded that when they carried out a logistic regressionanalysis, allowing for the grade of asbestosis, years of exposure accounted for most ofthe variation, but the degree of asbestosis still emerged as a highly significant riskfactor.

5.156 Recently, Case and Dufresne [190] have commented as follows:

WT/DS135/RPage 279

" ... Hughes and Weill go much further in stating that asbestosis is a prerequisite for lung cancerattribution in those with asbestos exposure. This statement goes beyond the known facts and relies onmechanistic speculation. The authors believe that asbestosis is produced by a mechanism ormechanisms that will also lead to lung cancer. Their hypothesis requires that the mechanism(s) alwaysbe intermediate in that lung cancer always follows asbestosis. Finally, the speculation requires thatlung cancer occurring without asbestosis can never be caused by asbestos exposure alone (or in synergywith cigarette smoking) regardless of the level of that exposure, and that no mechanism can occur thatdoes not involve intermediate fibrosis. The biological fallacy of this argument has been welldocumented ... one must remember that lung cancer originates in the large airways, while asbestosis isa disease of the lung parenchyma at and beyond the respiratory bronchioles. ... To ignore ourknowledge of indices of exposure other than the simple presence or absence of asbestosis is simplisticand biologically naïve." [p 1118].

5.157 The case-control studies carried out on South Carolina asbestos textile workers by Dement etal. [171] clearly undermine the fibrosis→cancer hypothesis and, in this respect, they constitutePopper's Black Swan factor:18 Dement and his colleagues clearly identified a lung cancer SMR > 2.5at 2.7-6.8 fibre-years of exposure (well below the exposure level necessary for histological asbestosisin the same cohort [191]).

(viii) Non-occupational asbestos exposure in Quebec and lung cancer risk

5.158 The first written submission from Canada also refers to the study reported by Camus et al.[140] on non-occupational exposure to chrysotile asbestos in Quebec and the risk of lung cancer:

"It is also interesting to note the work of Dr Camus et al. (see Camus, M., Siemiatycki, J. Meek, B.,Nonoccupational Exposure to Chrysotile Asbestos and the Risk of Lung Cancer, (1998) 338, NewEngland Journal of Medicine 1565). They published a vast study on women in chrysotile miningcommunities in Quebec, many of whom were exposed to very high levels of fibres between 1920 and1975. These women were subjected to exposure of 0.0107 f/ml,19 higher than the current exposurelimits in France, and literally thousands of times higher than the levels measured in public buildings.Nonetheless, no excess in lung cancer was detected in this population. According to the study'sauthors, this is particularly important in the light of the current French situation. In fact, applying therisk model adopted by France for the exposure studied, results in a forecast of approximately 100 lungcancer deaths, while in reality there are none. Likewise, use of the French risk model would haveresulted in estimates of approximately 250 and, at any rate, no less than 50 deaths from mesothelioma,while the preliminary results of the study in question show only 10 cases, some of which may beassociated with exposure to amphiboles. Research continues, particularly with an analysis of the workhistory of each individual in order to determine the exact link, if any, between these cases ofmesothelioma and on-the-job exposure, as well as exposure to amphiboles."

18"The philosopher of science, Sir Karl Popper ... coined the term 'falsification' to express the concept

that scientific theories are not proven by repetition of results but rather survive because they successfullywithstand refutation (falsification). His example of the black swan makes this point clearly. Suppose you havea hypothesis that all swans are white ... you observe, say, 10,000 swans and they are all white. Another scientistrepeats your efforts and observes another 10,000 swans: they too are all white. So far the theory is standing upwell. The repetition helped to strengthen it — but if only a single black swan is sighted, this falsifies the theory:it is no longer tenable. Popper asserted that scientific statements have to be formulated in a manner that subjectsthem to the possibility of falsification. One of the important demarcating criteria between science andnonscience, according to Popper, is this formulation of statements in a manner permitting falsification" [pp 18-19] [44].

19This figure is inconsistent with the former limit of 0.1 f/ml in France, and contradicts Case's claimthat the Quebec women were exposed at up to 1 f/ml [192]; at a level of 0.0107 f/ml (a figure two orders ofmagnitude less than 1f/ml), a cumulative exposure of 5 fibre-years would require residence of > 150 years(adjusted for equivalence to an 8-hour working day) and > 750 years to reach 25.0 fibre-years (using the sameadjustment).

WT/DS135/RPage 280

4.159 In fact, Camus et al. [140] investigated the relative risk of death from lung cancer among2242 deaths between 1970 and 1989 among women ≥ 30 years of age who lived in two chrysotileasbestos-mining areas that comprised eight towns of which three (Thetford mines, Black Lake andAsbestos) contained nearly all the asbestos mines and mills. Eighty percent of the women lived within4 km of a mine or mill, and all lived within 10 km.

5.160 The estimated average cumulative level of exposure was 25 fibre-years (range 5-125fibre-years) made up by neighbourhood exposure (16.0 fibre-years), household exposure of 7.8 fibre-years and occupational exposure of 1.2 fibre-years, making a total of 25.0. The authors of this studypointed out that:

" ... The lower limit of 5 fibre-years per ml corresponds, for example, to 50 years of exposure toasbestos at a level of 0.1 fibre per ml (the actual mean ambient airborne asbestos level in the area in1974); the upper limit of 125 corresponds for example, to 50 years of exposure to 2.5 fibres per ml — arelatively low exposure level in local asbestos-mining and asbestos-milling industries before 1960." [p1568].

5.161 This investigation found a standardized mortality ratio of 1.0 in comparison to the referencepopulation (i.e. no observed excess of lung cancer mortality). However, seven deaths from "pleuralcancer" were observed (RR = 7.64; p < 0.05).

5.162 A few points about this study are worth emphasis:

• The Quebec chrysotile miners and millers have a low risk of lung cancer incomparison to other cohorts, such as the South Carolina chrysotile textile workers, forwhom the frequency of lung cancer is at least 30 times higher. Therefore, it is notsurprising that the low risk of lung cancer in the chrysotile miners and millers ofQuebec extends across residents exposed environmentally to the same ore. In otherwords, the absence of a detectable increase in lung cancer mortality in femaleresidents of this region of Quebec may not apply to other groups exposedenvironmentally to asbestos from other asbestos industries.

• The study reported by Camus et al. [140] stimulated considerable correspondence inthe columns of the same journal (NEJM), and at least two of the correspondents(Churg [193] and Case [192]) emphasized that the seven-fold increase inmesothelioma mortality (seven cases) among the women was probably explicable byoccupational exposure to amphiboles from manufacture of gas masks, repair of bagsthat contained imported asbestos, and, possibly in one case, domestic exposure to"tremolite brought home on miners' clothes".

In his letter to the editor, Case [192] also pointed out that "[T]hese women wereexposed to levels of chrysotile as high as 1 fibre per ml of air as recently as onemonth in 1984."

• I have some misgivings over the exposure estimates for this female population, andthe figure of 25 fibre-years from environmental exposure in the generalneighbourhood or vicinity of the Quebec chrysotile industry seems high incomparison to neighbourhood or environmental exposures from other industries. Forexample, ECH 203 (p 35) reproduces a Table of asbestos fibre concentrations inQuebec chrysotile mining towns, where the fibre concentration in 1984 is in thevicinity of 0.005 fibre/ml and the concentrations in 1973 and 1974 are givenas 0.08 fibre/ml. In other words, Case's figure of 1 fibre/ml for one month in 1984[192] may be doubtful, unless there were some catastrophic event in the industry, with

WT/DS135/RPage 281

a burst of asbestos into the general environment. Unless earlier environmentalairborne fibre concentrations were substantially above the 1973/1974 concentrations,it is difficult to see how a cumulative exposure of 25 fibre-years would come about;e.g. Camus et al. [140] state that residence in the area for 50 years at a mean fibreconcentration of 0.1 fibre/ml would lead to the lower estimate of 5 fibre-years.

In addition, the estimate of 25 fibre-years seems high in comparison to data onenvironmental airborne fibre levels related to the Zimbabwean and Russian chrysotileindustries. For example, EHC (p 47) states:

"There are some data concerning fibre levels in the air close to chrysotile mines.Baloyi (1989) found fibre levels around the Shabani mine (Zimbabwe) to range frombelow the limit of detection of the method" (less than 0.01 f/ml) to 0.02 f/ml of air,assayed by PCOM. [PCOM = phase contrast optical microscopy].

Scherbakov et al. [163] also give a comparable environmental airborne fibreconcentration in Asbest City of 0.1 mg/m3 (comparative data for the same industry[194] suggest that the gravimetric measurement of mg/m3 is very roughly equivalentto the same number of fibres/ml).

The point is that if the estimate of cumulative asbestos exposure in the Quebec femalepopulation is high, this would lead to underestimation of lung cancer risk or mortality.For example, no detectable increase in lung cancer mortality among the 2242 deathswould be expected at the low cumulative estimate of ≤ 5 fibre-years.

• In addition, in their reply to the Letters to the Editor, Camus and Siemiatycki [141]state that "[W]e agree ... that the study had low statistical power to detect small risks;this was conveyed by the wide confidence intervals for our risk estimates ...",although they go on to indicate that the Quebec study should have detected a risk ofthe magnitude predicted by the Environmental Protection Agency [EPA].

(ix) The Helsinki Criteria

5.163 This set of criteria deals with attribution of lung cancer to asbestos for the individual patient[113]:

"Because of the high incidence of lung cancer in the general population, it is not possible to prove inprecise deterministic terms that asbestos is the causative factor for an individual patient, even whenasbestosis is present. However, attribution of causation requires reasonable medical certainty on aprobability basis that the agent (asbestos) has caused or contributed materially to the disease. Thelikelihood that asbestos exposure has made a substantial contribution increases when the exposureincreases. Cumulative exposure, on a probability basis, should thus be considered the main criterionfor the attribution of a substantial contribution by asbestos to lung cancer risk." [p 314; emphasis inoriginal].

5.164 The Helsinki Criteria set an exposure level of ≥ 25 fibre-years of exposure; however, itshould be emphasized that this level of cumulative exposure is required for the individual patient as anindex for an asbestos-attributable relative risk of lung cancer of ≥ 2.0 (which, in the individual patient,equates to a probability of causation or material contribution of ≥ 50 per cent — the civil standard ofproof). Intended as a criterion for individual compensation, this exercise is clearly different frompopulation-based relative risks relevant to the dispute before the WTO.

WT/DS135/RPage 282

5.165 In summary:

TABLE 10: ASBESTOS-RELATED DOSE-RESPONSE RELATIONSHIPS FOR LUNG CANCER

Chrysotile or Amphiboles

Heavy exposure Dose-response effect; linear

Low-level exposure Dose-response effect for South Carolina textile workers (chrysotile)

Threshold No threshold delineated

(j) Some General Observations on Experimental Models of Asbestos Carcinogenesis, includingin vivo and in vitro Systems

(i) In vivo experimental models

5.166 Although animal models of asbestos carcinogenesis - especially induction of mesotheliomasin animals such as rats - are of value to demonstrate the capacity of different fibres to induce tumoursand to elucidate the mechanisms underlying carcinogenesis, they are not strictly comparable tocarcinogenesis in humans, for a number of reasons:

• The airborne fibre concentrations to which experimental animals are exposed forinhalation experiments are substantially higher than in workplace or environmentalsituations for humans.

• The routes of administration of asbestos or other fibres - e.g. injection or directimplantation into the pleura or peritoneum - are not comparable to the humansituation, with the exception of inhalation experiments.

• High concentrations of asbestos or other fibres are necessary to reduce latencyintervals so that a reasonable yield of mesotheliomas or other cancers is obtainablewithin the life span of the animal used. In other words, the latency intervals are notcomparable to the human model.

• There are known to be marked differences in the susceptibility of different species toasbestos carcinogenesis.

5.167 For example, in a review of asbestos and lung cancer, Henderson et al. [131] state thefollowing:

"The dose of asbestos delivered by inhalation or installation over a short time interval in experimentalanimals, the lag-times, and the histological spectrum of the tumors also make it difficult or impossibleto extrapolate the findings from such models to humans. The exposure to asbestos in positiveinhalation experiments seems to have been so high that fibrosis was an unavoidable association with anincreased cancer risk (exposure to at least 100 f/ml, > 1,000 f/ml for some groups, 5 x 7 hours perweek, up to 12 months or more). Wagner et al. remarked on a number of 'surprising' results in theirstudy (e.g. no differences in carcinogenicity or fibrogenicity between chrysotile and the amphiboles). ...

The sensitivity of humans to the carcinogenic effects of asbestos is about 100-fold greater than that ofrats. ...

... Experimental studies of this type address asbestos inhalation in isolation, instead of asbestoscombined with tobacco smoke [for the study of lung cancer]. Hence, they are of questionable relevanceto most lung cancers in asbestos workers, for which tobacco smoke is an important co-factor.

WT/DS135/RPage 283

For the reasons stated above, we consider that the existing literature on tumorigenesis by inhalation ofasbestos in laboratory animals allows no conclusions on the asbestos-asbestosis-lung cancercontroversy in humans." [p 96].

5.168 Davis [195] comments in the following terms:

"In experimental inhalation and injection studies, however, chrysotile has repeatedly produced as manymesotheliomas as other asbestos types. This finding probably indicates that the carcinogenic potentialof chrysotile to cells is as high as the other asbestos types, and it is just sufficiently durable to exert itsmaximum effect in rats, although it is unable to survive long enough to do so in humans." [p 201; butsee discussion in this report on chrysotile clearance from lung tissue, Section A.(g)(v)].

(ii) In vitro systems

5.169 It is obvious that the effects of asbestos and other fibre types on isolated cell lines used forin vitro studies are not comparable to the induction of mesothelioma or lung cancer in humans. Invitro studies of this type are of most value in showing that asbestos and other fibres can inducechromosomal injury, oncogene expression or mutations similar to those induced by other knowncarcinogens.

5.170 Detailed discussion of the voluminous literature on this topic lies beyond the scope of thisreport. Henderson et al. [131] give some details of the effects of asbestos on cell lines in vitro; moreextensive reviews are given in EHC 203 (pp 69-102), Both et al. [196], and Mossman et al. [197-202],and Bielefeldt-Ohlmann [203]. Only a few recent studies on chrysotile follow:

• "In the study by Haugen et al. [204], chrysotile was about 10 times more cytotoxicthan amosite or crocidolite (as assayed by inhibition of clonal growth rate) and > 100-fold more toxic than glass fibres; epithelial cells were 10-15 times more sensitive tothe cytotoxic effects of asbestos fibres than bronchial fibroblasts from the samehuman. We can find no comparison with mesothelial cells in this paper [204], despiteat least one claim to this effect [197] ... " [p 97].

• "Harrison et al. demonstrated synergy between the lung carcinogenN-nitrosoheptamethyleneimine (NHMI) and chrysotile in the production ofhyperplastic epithelial lesions in the lungs of rats, with a dose-response relationshipfor NHMI, augmented by chrysotile. Neoplastic lesions (adenoma andadenocarcinoma) were found only in animals treated with both NHMI and asbestos,but the number of such tumours was small (N = 6 among 115 rats studied)" [p. 118;see Henderson et al. [131] for references].

• "Hei and Piao reported on malignant transformation of a human papillomavirus-immortalized human bronchial epithelial cell line (BEP2D) by a single 7-daytreatment with chrysotile: the cells so treated evolved through a series of sequentialsteps to become tumorigenic, with the formation of progressively growing tumours innude mice" [p. 118; see Henderson et al. [131] for references].

• In an investigation of the capacity of different asbestos fibre types to induce loss ofheterozygosity [LOH] mutations in lymphocytes and diploid mesothelioma cells thatwere heterozygous for the HLA A2/A3 histocompatibility complex studied (incollaboration with Dr David Turner at the Department of Haematology-Oncology atthe Flinders University), it was found that chrysotile was more toxic to the cell linesused so that few viable cells remained, making it difficult to evaluate LOH mutations,in contrast to UICC South African crocidolite.

WT/DS135/RPage 284

• More recently, Dr. Turner and I have investigated the effects of UICC South Africancrocidolite injected into the peritoneal cavity of mice, for the investigation of somaticintrachromosomal recombinational events in mice that are transgenic for the gene thatencodes the enzyme β-galactosidase; using PCR [the polymerase chain reaction], wedetected a 5-fold reduction in SIR within only a few days of administration of thecrocidolite. This finding parallels the results obtained with other carcinogens(e.g. cytotoxic drugs used for cancer chemotherapy) and may be explicable by areduction of SIR because the asbestos produces an increase in other classes ofmutation (e.g. point mutations or deletions), or because of impairment of DNA repairmechanisms.

5.171 The picture now emerging on asbestos carcinogenesis is a prolonged multistage parametricprocess [205], in which asbestos fibres may participate in both the initiation and promotion phases[196]. Some classes of mutation potentially inducible by asbestos - e.g. loss of heterozygositymutations - are implicated in the initiation or progression phases of cancer development in humans,thought to be related to loss of tumour suppressor genes (e.g. retinoblastoma, astrocytoma, andcolonic, gastric, prostatic and breast cancer) [206-211].

5.172 Free radicals — generated either from the surface of the fibres themselves [205, 212-215] orvia macrophages [213, 216-218] — have been shown to have genotoxic or clastogenic properties[205, 212-214, 217, 219, 220], and are also implicated in asbestos carcinogenesis.

2. Questions by the Panel and Comments by the Scientific Experts

Question 1:

High-density chrysotile products (i.e. products where chrysotile fibres are bound in amatrix, such as chrysotile-cement, as opposed to "friable" products, such as flocking and heatinsulation) represent the main use of chrysotile asbestos. The parties to this dispute disagree as tothe circumstances of exposure to chrysotile and the risks to human health associated with suchproducts. In this context, various questions arise with respect to the risks to human healthassociated with the use of high-density chrysotile products, in particular chrysotile-cement (ofparticular concern are installation, modification, repair, maintenance, demolition and disposal).

1.(a) Canada argues that workers who are at greatest risk of exposure to chrysotile asbestos are,in descending order: (i) chrysotile miners and workers employed in the processing (milling)industry; (ii) workers in the chrysotile textile industry; (iii) workers involved in the production offriction materials (such as brakes, clutches); (iv) workers involved in the manufacturing ofchrysotile-cement products; (v) workers involved in the removal of asbestos from buildings; and(vi) workers involved in construction, renovation, maintenance and the heat insulation of buildings.Furthermore, according to Canada, the last two categories are likely to be exposed to amphiboles.On the other hand, the European Communities argues that, in France, the secondary users, whichincludes installation, maintenance, repair, insulation, waste management and "handyman" typepersons, etc. are at the greatest risk of exposure and that they are mainly exposed to chrysotileasbestos, since, for some fifty years, chrysotile has represented about 97 per cent of asbestosconsumption in that country. Could you comment on these contrasting views, with a special focuson current uses and products?

Dr. de Klerk:

5.173 This question is rather curious and is either irrelevant or the wrong words have been used:trying to elucidate an ordering of working groups according to their "risk of exposure to chrysotileasbestos". The risk of an event is the probability that it will occur. The event in question here is that a

WT/DS135/RPage 285

worker will come into contact with chrysotile asbestos. It is certain that the workers in groups (i) to (v)are exposed to chrysotile so their risks are all the same and equal to 1.0. Workers in group (vi) maynot come into contact with chrysotile so their risk of exposure is less. The more relevant question hereis: who is likely to receive the most exposure and therefore have the greatest risk of disease? Ingeneral, workers in well-regulated industries, where government inspection is mandatory, where thereis a long history of efficient industrial hygiene practices, will have less risk of disease than those in thesmaller less well-regulated industries. A good example can be found with silicosis: the majority ofcases now occurring in both the USA and Australia arising from small unregulated industries with noawareness of risks or hygiene practices. Similar examples from own experience are: witnessing (in1992) use of Russian asbestos in an asbestos-cement factory in Czechoslovakia (as it was then) whereall the warnings on the bags of asbestos were in English; passing by demolition in progress of an oldasbestos cement factory building in Sydney last month where no observable precautions of any kindwere being taken. (Note added later: It has struck me that the misunderstanding with this questioncould be due to the relative imprecision of the French language, where "de" means both "of" and"from", words with quite different meanings, especially in this context!)

Dr. Henderson:

5.174 In past historical terms, the Canadian proposition about the classes of workers at risk ofexposure to chrysotile is correct — provided that this risk is expressed in terms of a numerical valuefor the risk per person-years of observation (e.g. per 100,000 or 1 million person-years). However,this situation has changed over recent years, as airborne fibre concentrations have been reduced in themining and milling industries and during the production of friction products. As one example,NICNAS 99 points out that manufacture of friction products (brake linings and gaskets) in Australia isa completely closed operation, with low airborne fibre concentrations.

5.175 EHC 203 refers to this reduction in airborne fibre concentrations:

"Based on data mainly from North America, Europe and Japan, in most production sectors workplaceexposures in the early 1930s were very high. Levels dropped considerably to the late 1970s and havedeclined substantially to present day values. In the mining and milling industry in Quebec, the averagefibre concentration in air often exceeded 20 fibres/ml (f/ml) in the 1970s, while they are now generallywell below 1 f/ml. In the production of asbestos-cement in Japan, typical mean concentrationswere 2.5-9.5 f/ml in 1970s, while mean concentration of 0.05-0.45 f/ml were reported in 1992. Inasbestos textile manufacture in Japan, mean concentrations were between 2.6 and 12.8 f/ml in theperiod between 1970 and 1975, and 0.1-0.2 f/ml in the period between 1984 and 1986. Trends havebeen similar in the production of friction materials: based on data available from the same country,mean concentrations of 10-35 f/ml were measured in the period between 1970 and 1975, whilelevels 0.2-5.5 f/ml were reported in the period between 1984 and 1986. In a plant in theUnited Kingdom in which a large mortality study was conducted, concentrations were generally above20 f/ml in the period before 1931 and generally below 1 f/ml during 1970-1979." [pp 2-3].

5.176 In contrast, the risk per million person-years of observation may be less in buildingconstruction, renovation and maintenance workers, but this smaller risk is spread across a substantiallylarger workforce (i.e. there are many more carpenters/joiners, builder's labourers, electricians,plumbers and other tradespeople in Western societies than the numbers of workers engaged in themining, milling or production of high-density asbestos-containing materials such as asbestos-cementsheets and pipes or brake blocks).

5.177 According to EHC 203:

"It should be recognized that although the epidemiological studies of chrysotile-exposed workers havebeen primarily limited to the mining and milling, and manufacturing sector, there is evidence, based onthe historic pattern of disease associated with exposure to mixed fibre types in western countries, that

WT/DS135/RPage 286

risks are likely to be greater among workers in construction and possibly other user industries."[EHC 203, p 9].

"Past uncontrolled mixed exposure to chrysotile and amphiboles has caused considerable disease andmortality in Europe and North America. Moreover, historical experience to mixed fibre types inEuropean countries has clearly indicated that a larger proportion of mesotheliomas occurs in theconstruction trades than in production. Far larger quantities of chrysotile than of other types of asbestoswere used in most construction applications. Epidemiological studies that contribute to ourunderstanding of the health effects of chrysotile conducted to date and reviewed in this monographhave been on populations mainly in the mining or manufacturing sectors and not in construction orother user industries. This should be borne in mind when considering potential risks associated withexposure to chrysotile." [EHC 203, p 137].

"Few data on concentrations of fibres associated with the installation and use of chrysotile-containingproducts were available to the Task Group, although this is easily the most likely place for workers tobe exposed." [EHC 203, p 138].

"There is potential for widespread exposure of maintenance personnel to mixed asbestos fibre types dueto the large quantities of friable asbestos materials still in place. In buildings where there are controlplans, personal exposure of building maintenance personnel in the USA, expressed as 8-h time-weighted averages, was between 0.002 and 0.02 f/ml. These values are the same order of magnitude asexposures reported during telecommunication switch work (0.009 f/ml) and above-ceiling work(0.037 f/ml), although higher concentrations have been reported in utility space work (0.5 f/ml).Concentrations may be considerably higher where control plans have not been introduced. Forexample, in one case, short-term episodic concentrations ranged from 1.6 f/ml during sweepingto 15.5 f/ml during cleaning (dusting off) of library books in a building with a very friable chrysotile-containing surface formulation. Most other values, presented as 8-h timed-weighted averages, areabout two order of magnitude less." [EHC 203, p 139].

5.178 These points are also borne out by the 1999 Report for the Australian Mesothelioma Register[AMR 99], where the broad spread of prior occupations among mesothelioma victims is plain. Forexample, the number of mesotheliomas from the former Wittenoom blue asbestos industry(189 mesotheliomas related to a single exposure only; 25 additional mesotheliomas as a consequenceof multiple exposures; total = 214) is less than the numbers of mesotheliomas as a consequence ofasbestos exposure in different occupations (e.g. carpenters/joiners: 187 mesotheliomas from a singleexposure; 33 additional mesotheliomas due to multiple exposures; total = 220; for builders/builders'labourers the corresponding numbers are 150 + 27 = 177). In other words, mesotheliomas among theformer Wittenoom cohort constitute a relatively small number (214) in comparison to the aggregatenumbers of mesotheliomas from asbestos exposures in other occupations (2585 – 214 = 2371 otherasbestos-associated mesotheliomas; no exposure data for 717 cases, and no apparent exposurefor 443; aggregate total = 3745).

5.179 NICNAS 99 makes the same point (p 59):

"Occupation/industry classification of the mesothelioma cases on the register are based on theAustralian Bureau of Statistics 'Industry and Occupation Codes'. The percentage of overall cases ofmesothelioma (January 1986 to March 1995) according to exposure category are: repair andmaintenance of asbestos material (13%), shipbuilding (3%), asbestos cement production (4%), railways(3%), power stations (3%), boilermaking (3%), mining (Wittenoom) (5%), wharf labour (2%), para-occupational, hobby, environmental (4%) carpentry (4%), building (6%) navy (3%), plumbing (2%)brake linings (manufacture/repair) (2%) and combinations of the above (multiple) (12%) (Leigh et al.,1997). Leigh (1994) reported that the pattern of exposure is shifting away from the older traditionalindustries towards product, domestic and environmental exposure. An analysis of 16 years data in 1996by Yeung et al. (1997) showed more cases (on a number of cases basis) in more recent years in theasbestos user industries and from occupations such as plumbers, carpenters, machinists and carmechanics."

WT/DS135/RPage 287

5.180 Similar patterns of exposure — and resultant diseases (lung cancer; mesothelioma) — havebeen recorded in the United Kingdom (EHC 203, pp 123-124):

"Based on analyses of mortality of workers with mixed exposures to chrysotile and amphiboles in theUnited Kingdom, by far the greatest proportion of mesotheliomas occurs in users of asbestos-containing products, rather than those involved in their production. ...

1. Asbestos exposure caused approximately equal numbers of excess deaths from lung cancer(749 observed, 549 expected) and mesothelioma (183 deaths) within the occupations covered by the1969 and 1984 Regulations ...

2. Only a few (5%) of British mesothelioma deaths were among workers in regulatedoccupations (Peto et al., 1995). The majority of deaths occurred in unregulated occupations in whichasbestos-containing products are used, particularly in the construction industry. The risk wasparticularly high among electricians plumbers and carpenters as well as among building workers."

5.181 As shown by the literature cited in this discussion, it is my perception that there is broadagreement among experts on these patterns of exposure.

Dr. Infante:

5.182 The relative exposure categorization of the six job situations mentioned in the questiondepends on the nature of controls being used in each situation. In general, exposures are more easilycontrolled in manufacturing and more difficult to control in construction, maintenance, repair,demolition and disposal activities. Today, exposures would be more easily controlled in mining andmilling because of awareness of the hazard and the clear identification of the operations as sources ofasbestos exposure. Quite often workers involved in maintenance, repair and handyman type activitiesdo not know whether asbestos is present or not. In the absence of such knowledge, workers usually dolittle, or nothing to protect themselves from exposures to asbestos in these situations. As a result,workers involved in these activities are most likely to be the most heavily exposed in the occupationalsetting today. These types of activities often result in asbestos being carried home on the workers'clothing. A typical scenario that comes to mind is a situation whereby a worker is in a crawl spaceand encounters asbestos insulation. There is no active supervision in this situation and the asbestosmost likely is not labelled. Thus, the worker cuts through the insulation to get to the area needing tobe repaired without knowledge of the hazard and without having the appropriate personal protectiveequipment. In the latter scenario, even when workers do wear respirators, they are often dust masks,which do not provide a proper face seal and the filter medium is not adequate i.e., HEPA filters are notpart of the filtration material on these masks. In the repair trades particularly, it is common practice forworkers to use dust masks that do not provide HEPA filtration. As a result, the respirator isinadequate for filtering out the fibres of dimensions that are thought to lead to cancer and otherasbestos related diseases. Furthermore, even in situations where the appropriate respirators may beworn, comprehensive respiratory fit-testing programmes may not be included as part of the industrialhygiene programme and as a result, the respirators leak because of the inability to achieve a properface seal. In situations where workers may be drilling, sawing, crushing, or sanding asbestos cementproducts the only appropriate respirator may be a supplied air respirator, but it may not be usedbecause it is too cumbersome for the job situation. In my opinion, scenarios (v) and (vi) are usuallythe most dangerous in current times because the workers are not aware of the presence of asbestos andthey are more likely not to have received training and education about the hazards of asbestosexposure.

5.183 The risk of exposure should be considered not only by level of exposure, but also by theextent of the populations exposed to chrysotile asbestos. The large number of mesotheliomasassociated with secondary and tertiary users of chrysotile asbestos (maintenance workers, electricians,bystanders, etc.) is a reflection of the large number of individuals in the population exposed in these

WT/DS135/RPage 288

situations. Thus, in terms of the risk of disease from chrysotile exposure, one must consider not onlythe intensity of exposure in the various work situations, but also the extent of the population exposed.One study (Begin et al., 1992) reports that 33 per cent of mesothelioma cases identified amongmaintenance workers, electricians, bystanders, etc. were the result of exposure for less than five years,and that the incidence of these occasionally exposed cases was increasing more rapidly than in theprimary industries (mines and mills), or in the secondary industries (manufacturing, daily handling ofasbestos).

Dr. Musk:

5.184 The term "risk of exposure" is taken to mean who is most likely to receive the most exposureand therefore be at the greatest risk of developing asbestos-related disease. This would depend on thenature of the industry in the locality and the type of asbestos being produced or used or otherwiseencountered. Those workers likely to receive the most exposure would be those in industries whereregulations are most permissive or compliance with them is poorest from absence of supervision ormeans of personal protection. It would also depend on the conditions of work such as indoor versusoutdoor etc. Canada's "argument" could be settled by monitoring of exposure! The "arguments" donot seem to be incompatible.

1.(b) Should we consider that the risk to human health associated with the various uses ofchrysotile throughout its life-cycle is a workplace issue or does this risk affect a larger part of thepopulation?

Dr. de Klerk:

5.185 The risk of disease from chrysotile affects everyone. The risk of disease depends on intensityof exposure, the duration of exposure and the time since exposure. The population who do not workwith asbestos will still come into contact with it, albeit at a much lower intensity, however thispopulation is much larger and hence the burden of disease may be greater. There are numerousexamples of asbestos-related disease arising in people living in the vicinity of asbestos works or livingwith asbestos workers.

Dr. Henderson:

5.186 From my perspective, this is overwhelmingly a workplace issue (e.g. construction workers).The risk of cancer for the larger general population from exposure to asbestos in place has beendiscussed in an earlier part of this report (see above section C.1.(e)). Please see also my answer to thepreceding question.

5.187 For example, asbestos-cement roofs are common in Germany where corrosion by acid rainrepresents a potential problem. Measurements carried out by Spurny et al. [221-224] on airborneasbestos fibre concentrations in the vicinity of such buildings consistently reveal levels in the order of0.0002-0.0012 f/ml, in comparison to fibre concentration in other urban environments that range upto 0.1 f/ml (but generally ≤ 0.001 f/ml).

5.188 Measurements have also been made on airborne fibre levels related to asbestos-cementroofing in schools in Western Australia [128] , with only one asbestos fibre detected in each of twoschools (air monitoring at 9 sites over 720 hours). Based on the findings, it was estimated thatairborne fibre concentrations would be unlikely to exceed 0.002 f/ml and were likely tobe < 0.0002 f/ml. These levels were considered to represent a negligible risk to health; theWestern Australia Advisory Committee on Hazardous Substances that carried out this investigationconsidered that a greater risk to health would arise from: (i) unskilled attempts to clean up the

WT/DS135/RPage 289

asbestos-cement roofs before application of protective coating; and (ii) trauma to the workers —e.g. falling from or through the roofs.

5.189 It is my perception that there is little or no dispute among experts on this issue.

Dr. Infante:

5.190 In general, workers are at relatively greater risk of exposure to chrysotile and disease,particularly those involved in maintenance, modification, demolition, repair and disposal activities ascompared to those exposed in non-occupational situations. A large number of people from the generalpopulation, however, also will be exposed to chrysotile and elevated risk of disease when they engagein home repairs that involve manipulating or disturbing asbestos-containing products. (The latterindividuals usually have little or no education about the hazards of asbestos, nor of the mostappropriate means to handle it with the least amount of exposure.) These types of operations will alsocreate some standby exposures (Ascoli et al. 1996). If appropriate controls are not used whenhandling asbestos insulation in buildings, the building can become contaminated and the occupantswill become exposed. Therefore, the major problem with asbestos exposure is related to occupationalsituations though a much larger population is exposed beyond the occupational setting to relativelylower levels. Reports of cases of mesothelioma among non-occupationally exposed individualsdocument non-occupational exposures to asbestos causing disease. Family members of workersinvolved in the asbestos cement industry (Magnani et al. 1993) as well as children of miners andmillers (McDonald and McDonald 1980) have been diagnosed with mesothelioma.

Dr. Musk:

5.191 It is my opinion that the risks resulting from exposure affects all exposed people and dependson the cumulative level of exposure. It is also my opinion that there is not an exposure thresholdbelow which there is no risk. The risks to people not occupationally exposed to asbestos are likely tobe much less than the risks to those with occupational exposure because the degree of exposure islikely to be less (though not necessarily always so). However, while the individual risks may be muchless the total burden of disease in the community may not be because it is likely that there are manymore people experiencing these risks (albeit lower). For example the burden of disease in theresidents of the town of Wittenoom, Western Australia has been significant albeit less than that of theworkers. The WA Mesothelioma Registry contains subjects whose only exposure was fromneighbourhood industries. Similar cases have been documented in the Quebec areas.

1.(c) Can chrysotile-cement products (for instance in buildings) release fibres, throughweathering, corrosion or general degradation, thus presenting a possible risk to human health?Can you quantify this risk?

Dr. de Klerk:

5.192 There is good evidence that both wind and rain cause the release of fibres even from newasbestos cement sheeting. Other possibilities are fires and unwanted demolition. It is hard to quantifythe risk which again depends on intensity and duration, but measurements have been made in thevicinity of such buildings which are detectable but low.

Dr. Henderson:

5.193 Please see the preceding answer, and section C.1.1(e). Quantitation of the risk is based onbackward extrapolation according to the linear no-threshold model because there are no observationaldata on the dose-response effects from low-level exposure to chrysotile, and the estimates are,

WT/DS135/RPage 290

therefore, open to question and dispute, but the risks to health from very low-level environmentalexposure appear to be minuscule or negligible.

Dr. Infante:

5.194 Yes, weathered and corroded asbestos cement products are capable of releasing chrysotilefibres into the environment, and most of the fibre is transported by rainwater though some will bereleased into the ambient air in low concentrations. One study indicates that chrysotile exposure insuch circumstances will generally be less than 1,000 fibres longer than 5 microns per cubic meter ofair. The fibres released were shown to have the same carcinogenic potency as "standard" chrysotilefibres (Spurny, 1989). Asbestos fibres also will be released into water from cement water pipes.I have not seen any estimates of risk from this type of asbestos exposure. Although the relative risk ofdisease is considerably less than that from occupational exposures, the population at risk isconsiderably larger.

Dr. Musk:

5.195 I understand that asbestos-cement products do release fibres as they weather. Release offibres occurs from both old and new products. Asbestos fibres can also be released when asbestos-cement products are involved in fires. Quantitative estimates of the risks are theoretically possible asairborne concentrations can be measured and dose-response relationships are known.

1.(d) Can interventions on chrysotile-cement and other high-density chrysotile products releasefibres, thus presenting a possible risk to the health of the individual making such interventions or tothe public in general? Can you quantify this risk?

Dr. de Klerk:

5.196 It is during interventions such as drilling, sawing, sanding, moving in stacks, loading ontotransport etc, that concentrations of fibres are greatest, both for the operators and bystanders. Theconcentrations associated with such operations have been extensively tabulated in the literature.Exposure response relationships can be used to estimate the risk for any combinations of intensity,duration and time after exposure, as shown in the Table below.

5.197 Lifetime risks (to age 85 years) of mesothelioma after exposure to chrysotile,assuming 0.1 f/ml for 10 years from age 20 with competing causes of death at 1992Western Australian death rates.

Assumptions Expected cases per million lifetimes

Health Effects Institute equation 724

Wittenoom crocidolite equation 1/12th potency 210

Wittenoom crocidolite equation , 1/80th potency 32

Background risk (Peto study of "unexposed"Los Angeles population)

112

Dr. Henderson:

5.198 My answer to the first question is YES. Operations such as drilling or sawing asbestos-cement products release fibres and produce elevated airborne fibre concentrations. (i) asbestos-cement sheets can release respirable fibres in the absence of manipulation, even when new (upto 0.001 f/ml; for references, see de Klerk and Armstrong [135]); (ii) a 1938 report in

WT/DS135/RPage 291

New South Wales indicated that cutting asbestos-cement products with a power saw couldgenerate 4-5 million particles/cubic foot (roughly equivalent to 12-15 f/ml); cutting with hand sawsproduced lower concentrations; (iii) as shown in the following Table 11, Sturm et al. [5, 7] reportedmeasurable fibre concentrations from various operations on asbestos-containing materials, includingasbestos-cement in the former East Germany, as measured by occupational inspectors.

TABLE 11: ASBESTOS FIBRE CONCENTRATIONS AT WORKPLACES, WITHOUT SUCTION DEVICES,DETERMINED BY KONIMETRY (FROM UNPUBLISHED REPORTS

PREPARED BY OCCUPATIONAL INSPECTORATES)

Type of Work Fibre Concentration (f/ml)

Scratching and crushing of asbestos-cement 0.03 to 0.3

Abrasive cutting of asbestos-cement without dust removal by suction 0.3 to 10.0 approx.

Drilling asbestos-cement without dust removal by suction 0.5 to 3.4

Machining of brake linings 0.1 to 13.0

Replacement of gaskets 0.02 to 0.5

Punching of gaskets (rubber asbestos) 0.02 to 1.9

Use of asbestos gloves 0.02 to 0.6

Replacement of clearing layers 0.06 to 0.5

Use of talcum for powdering gloves 0.6 to 20.0

Level limit value (over a whole working day) 1.0

5.199 In 1993, Kumagai et al. [4] in Japan reported on dust levels generated by repair work onasbestos-cement pipes, including use of a high-speed disc cutter both inside holes dug in the ground togain access to the pipes and outside the holes. The concentration of asbestos fibres > 5 µm in lengthranged from 48-170 f/ml inside the hole (average = 92 f/ml) and ranged from 1.7-15 f/ml outside thehole. The Abstract for this paper follows:

"Asbestos cement pipes (ACPs) containing 15 to 20% chrysotile or crocidolite have been used forunderground conduits. Even today 16.2% of all conduits in Japan are ACPs, though the production ofACPs was suspended in 1985. When such a conduit is accidentally damaged the workers belonging tothe Waterworks Bureau of a local government cut off the damaged conduit using a high-speed diskcutter and replace it with a new conduit. This operation develops a cloud of dust and the workersinvolved run the risk of asbestos exposure. It was the aim of the present study to estimate asbestosexposure levels among these workers. First, in the experiment, we established the typical workingconditions and requested an experienced worker to cut an ACP using a high-speed disc cutter in a holedug in the ground as he routinely does. The experiment was repeated three times. During a bout ofeach experiment, dust was sampled at several points both inside and outside the hole. Second, a self-administered questionnaire survey was conducted to obtain information from the workers regardingtheir working conditions in cutting ACPs. The subjects of the survey were 1,048 men belonging toconduit repair sections of the Waterworks Bureau of 119 local governments. The results obtained canbe summarized as follows. (1) Each bout of cutting ACPs required about five minutes. Theconcentration of asbestos fibers longer than 5 microns with 3:1 aspect ratio ranged from 48to 170 fibers/ml (92 fibers/ml on an average) inside and 1.7 to 15 fibers/ml outside the hole. Theconcentration inside the hole exceeded the ceiling limit (10 fibers/ml) recommended for asbestos by theJapanese Association of Industrial Health. A concentration of 92 fibers/ml is equivalentto 0.96 fibers/ml as 8-h time-weighted average. (2) The number of subjects with experience of cuttingACPs was 849 (81.0%). The average length of service in conduit repair section was 14.2 yr. Based onthe information obtained from each subject regarding the average working days per yr for each decadefrom 1946, the cumulative days to date expended in cutting ACPs was estimated to average 235 d, thatis, 17 d per yr. Only 18.1% of the subjects used a protective respiratory device."

WT/DS135/RPage 292

5.200 EHC 203 also gives the following data (p 40):

"Weiner et al. (1994) reported concentrations in a South African workshop in which chrysotileasbestos-cement sheets were cut into components for insulation. The sheets were cut manually, sandedand subsequently assembled. Initial sampling showed personal sample mean concentration of 1.9 f/mlfor assembling, 5.7 f/ml for sweeping, 8.6 f/ml for drilling and 27.5 f/ml for sanding. Afterimprovements and clean-up of the work environment, the concentrations were 0.5-1.7 f/ml.

Nicholson (1978) reported concentrations of 0.33-1.47 f/ml in a room during and after sawing andhammering of an asbestos-cement panel."

5.201 It is my perception that there is no dispute among experts on this issue.

5.202 In relation to the second part of this question, apart from stating that there is a risk because ofthe generation of airborne asbestos fibres from interventions on asbestos-cement and other high-density asbestos products, it is not possible to quantify the risk in a way that would meet withuniversal agreement or a broad consensus, because few data are available for the risks for this type ofoperation on chrysotile-cement products: the risk would be related to cumulative exposure, whichwould vary according to the types of operation carried out, and their frequency. In addition, riskestimates would be dependent on extrapolation from the linear dose-response model that has beencalled into question by Canada. Therefore, I would expect disagreement among authorities on themagnitude of the risk.

5.203 Table 12 derived from NICNAS 99 gives risk estimates for lung cancer at airborne chrysotileconcentrations of 0.1-1.0 f/ml, according to the National Occupational Health and Safety Commissionin Australia (NOHSC) and two US occupational health and safety bodies (OSHA and NIOSH).

TABLE 12: ESTIMATED RISK OF LUNG CANCER AT VARIOUS LEVELS OF EXPOSURE TO CHRYSOTILE

Exposure(yearly average fibre/ml) Excess risk (per 100,000 persons exposed)NOHSC US OSHA US NIOSH

1 173 2880 57600.5 86 1440 28800.1 17 288 576

Excess risk = Risk coefficient x lifetime exposure (yrs) x average exposure level (f/ml) background risk.**A cumulative background risk for lung cancer in the male population was used in these calculations(i.e. 7200/100,000 assuming mixed smoking habits).

5.204 However, NICNAS 99 goes on to discuss uncertainties concerning these risk estimates:

"There are several other reasons why there is considerable uncertainty regarding these risk estimates,which include:

1. Past occupational exposures have generally involved exposure to a mixture of asbestos fibres. As itappears likely that different types of asbestos have different degrees of hazard, it is difficult todetermine the risk attributable to chrysotile per se. In addition, commercial chrysotile often haslow levels of tremolite contamination.

2. Fibre size, such as difference in fibre size between different chrysotile industries, probablyinfluences the degree of hazard and/or potency.

3. There is a long latency between exposure to asbestos and development of lung cancer. Hence, it isnot possible to state definitively what fibre type and level of exposure caused the disease.Consequently, risk estimates are related more to duration of employment rather than intensity ofexposure.

WT/DS135/RPage 293

4. A linear, non-threshold model may not be an appropriate model as there is some evidencesuggesting that lung cancer due to chrysotile exposure may have a threshold for effect.

5. Past exposure estimates (both quantitative and qualitative) are subject to considerable error. Forexample, conversion of historical results in mpcf units to fibres/mL has inherent uncertainties.

6. There is a high background level of lung cancer in the general population due to smoking. Cases oflung cancer attributable to asbestos cannot be distinguished from those due to smoking. Attributioncan only be assessed in terms of excess of lung cancers above a control population, hence the choiceof control population is critical.

7. The identification of the disease is dependent on medical diagnosis, however autopsies are notalways conducted.

The impact of some of these uncertainties can be accounted for to some extent. For example, it isconsidered that (1) and (2) are largely accounted for by basing risk estimates on epidemiological studieswhere exposure was only to chrysotile in the most relevant industry. For the remainder of the aboveuncertainties it is unclear what influence they have on the risk estimates and how they should beaccounted for. For example, recently there has been some debate in the literature as to whether athreshold or non-threshold model should be used when predicting risk due to chrysotile exposure.Meldrum (1996) states that based on balance of toxicological evidence, the linear no-threshold modelfor chrysotile-induced lung cancer may not be appropriate. ... Epidemiological data alone are not ableto clearly distinguish between the possibility of a threshold or a non-threshold model due to therelatively high background rate of lung cancer in the human population. There is at present noconsensus with respect to a threshold level of exposure for chrysotile below which there is no risk ofdisease" [pp 70-71].

5.205 Table 13 gives an estimate of lifetime mesothelioma risk from exposure to low levels ofchrysotile (1.0 f/ml and 0.1 f/ml), based on dose-response data for the Wittenoom cohort, andassuming lower potencies for chrysotile than crocidolite (i.e. 1/12th, 1/30th and 1/80th).20

TABLE 13: ESTIMATES OF LIKELY MESOTHELIOMAS RELATED TO CHRYSOTILE INHALATION ATAIRBORNE CONCENTRATIONS OF 1.0 AND 0.1 F/ML, ASSUMING A CARCINOGENIC

POTENCY 1/12TH, 1/30TH OR 1/80TH THAT OF CROCIDOLITE

Numbers to age 85Mesotheliomas/million persons

Airborne fibre concentration; Potency Duration of exposure* (years)

1 yr 10 yrs 40 yrs1.0 f/ml -1/12th 282 2101 35301.0 f/ml - 1/30th 113 840 14120.1 f/ml - 1/12th 28 210 3530.1 f/ml - 1/30th 11 84 1410.1 f/ml - 1/80th 4 32 53

*Starting at age 20 yrs.

20Estimates calculated at my request by Dr. N.H de Klerk.

WT/DS135/RPage 294

Dr. Infante:

5.206 Interventions on chrysotile cement products can result in extremely high atmospheric fibreconcentrations (Rodelsperger et al., 1980) and studies of roofers have demonstrated asbestosis fromsuch exposures (Stauder et al., 1982). A study of workers involved in inside finishing work withconcrete asbestos containing 30 per cent chrysotile asbestos has also shown that airflow obstructionamong workers can be caused by such exposure (Harless et al. 1978). Open-air cutting such as thatinvolved in roofing and inside finishing operations also will result in exposure to other workers notdirectly involved in the manipulation of the asbestos, e.g., standby exposure. Such manipulation ofasbestos cement will also expose the general population.

5.207 A study of 404 roofers with long term exposure to cement dust indicated that 14 per cent hadsignificantly increased small irregular opacities with profusion in 13 per cent (Stauder et al., 1982).The prevalence of these abnormalities was significantly greater than that observed in the controlgroup. The study by Harless et al. (1978) indicated that approximately 50 per cent of workersexposed for about six months to dust from asbestos cement developed airflow obstruction. The risk ofdeveloping lung pathology per unit of asbestos fibre exposure cannot be determined from these studiesbecause of lack of exposure data. The studies do indicate, however, that uncontrolled manipulation ofchrysotile cement products can result in a high rate of lung pathology. Lung function can be adverselyaffected as a result of exposure over a very short period of time.

5.208 A large number of reports indicate mesothelioma related to car mechanics involved in brakerepair. General population exposures from such work would be minimal except for those situationswhere individuals would engage in their own brake repair work.

Dr. Musk:

5.209 Interventions on asbestos-cement products can release fibres therefore a risk of disease existsas in 1(c).

1.(e) Can occasional interventions on high-density chrysotile products, either in occupationalcircumstances (such as electricians, plumbers, repairers, insulation workers, etc.) or by privateindividuals ("handyman" type) release fibres, thus presenting a possible risk to the individualmaking such interventions or to the public in general? Can you quantify this risk?

Dr. de Klerk:

5.210 Yes of course, see (c) and (d).

Dr. Henderson:

5.211 To answer the second question first, I am unable to quantify potential risk, because there areno systematic observational data available for this type of work, to the best of my knowledge (butplease see Tables 12 and 13 above in my answer to Question 1(d)

5.212 The first part of the question has been covered in the preceding answer, with the observationthat occasional interventions of this type would predictably produce low cumulative exposures, with alower risk, for the reasons discussed earlier. Please also refer to AMR 99, for data on mesotheliomasamong electricians, carpenters, plumbers, insulation workers and so forth (it is acknowledged thatmost if not all these mesotheliomas are a consequence of exposure to asbestos-containing materialsthat included a mixture of asbestos types, including chrysotile and one or more of the amphiboles); indrawing attention to AMR 99, my purpose is simply to use mesothelioma rates as a reflection of pastexposures and hence evidence that airborne fibre concentrations were produced by these types of

WT/DS135/RPage 295

operation, without regard to the fibre types. My own cases of mesothelioma also include a number ofindividuals whose only exposure to asbestos took the form of maintenance work and renovationscarried out on the patient's home, where there were asbestos-cement building materials. Again, indrawing attention to this type of background, it is not my intention to address fibre type, but simply toindicate that mesothelioma as an outcome of this type of exposure indicates that elevatedconcentrations of respirable airborne fibres were produced.

5.213 EHC 203 gives the following account (pp 122-123):

"Although the odds ratio for lung cancer associated with exposure to "asbestos" has been estimated inmany case-control studies, the studies have not been in general able to distinguish between chrysotileand amphibole exposure, and are therefore less informative for the present evaluation ... In a multisitecase-control study from Montreal, Canada, however, exposures to chrysotile and to amphiboles wereseparated, although exposure to amphiboles was not controlled for in the analysis for exposure tochrysotile (Siemiatycki, 1991). In this study, the occupational history of male cases (age 35-70) ofcancer at 20 sites and of 533 population controls was evaluated by a team of industrial hygienists andchemists to assess exposure to 293 agents. Overall, the lifetime prevalence of exposure to chrysotilewas 17% and that of exposure to amphiboles 6%. The main occupations involving exposure tochrysotile that were considered were motor vehicle mechanics, welders and flame cutters, andstationary engineers. When lung cancer cases (N = 857) were compared with cases of all other types ofcancer, the odds ratio (OR) of any exposure to chrysotile was 1.2 (90% CI = 1.0-1.5; 175 exposedcases), and that of 10 or more years of exposure with at least 5 years of latency ('substantial exposure')was 1.9 (90% CI = 1.1-3.2; 30 exposed cases). Corresponding ORs of exposure to amphiboles were1.0 and 0.9. The OR of exposure to chrysotile was higher for oat cell carcinoma than for other types oflung cancer. Twelve cases of mesothelioma were included in this study. The OR of any exposure tochrysotile was 4.4 (90% CI = 1.6-11.9; 5 exposed cases) and that of substantial exposure was 14.6(90% CI = 3.5-60.5; 2 cases). Corresponding ORs of exposure to amphiboles were 7.2 (90% CI = 2.6-19.9; 4 cases) and 51.6 (90% CI = 12.3-99.9; 2 cases). "

5.214 Please see also Tables 5, 9, 10, 11, 12, 13, in EHC 203.

5.215 It is my perception that there is no dispute among experts that such interventions releasefibres; disagreement is likely over the magnitude of the risk.

Dr. Infante:

5.216 The worker making the intervention would be the most highly exposed and presented with thegreatest risk of asbestos related diseases. The extent of exposure to the worker as well as to those inthe surrounding area would depend on the nature of the intervention, e.g., the circumstances underwhich the chrysotile asbestos product is manipulated in terms of work practices, the controls, or lackof controls in place and the type of personal protective equipment provided to the worker. While dataon fibre exposure levels in these situations are sparse, data on mesothelioma indicates association withworkers who have jobs that result in occasional interventions to asbestos products. Because theseexposures are not routine and the hazard often goes unrecognized, these operations are unlikely to bewell controlled, i.e., they are not anticipated so proper training and education about these types ofexposures is often lacking.

5.217 It is difficult to quantify this risk because atmospheric measurements are usually not madeduring these interventions. However, the identification of cases of mesothelioma associated withthese interventions in the literature indicate that they are perhaps the most detrimental to humanhealth. Mesothelioma has been identified from these exposure situations because it is a marker cancerrelated to asbestos exposure. What goes unidentified and unmeasured from these situations is themuch larger burden of disease and death from pneumoconiosis and lung cancer. The attributableburden from these latter diseases will be much greater than that from mesothelioma, but they are notusually recognized because lung cancer has a high background rate in the general population and

WT/DS135/RPage 296

asbestosis may be diagnosed as another type of pneumoconiosis unrelated to asbestos exposure.Mesothelioma has also been documented among the wives of construction workers, indicating that thefamily member portion of the public in general is also at risk. These latter cases of mesothelioma aremost likely the result of carry-home exposure from contaminated clothing.

5.218 If one considers that the handyman types of exposures are considered as exposures to thegeneral public, then this segment of the general population would also be at an elevated risk ofdeveloping asbestos related diseases. Exposures to family members that might result frominterventions by home owners would depend on the nature and location of the removal ormanipulation of the asbestos. The general public is also exposed through manipulation of asbestos inresidential buildings that is not carried out with appropriate controls in place and by carry homeexposures from contaminated work clothing.

Dr. Musk:

5.219 Occasional interventions on asbestos-cement products by anyone can release airborne fibres,therefore there is some risk as in Question 1(c).

1.(f) Are chrysotile fibres from chrysotile-cement dust released during interventions (cutting,sawing, etc.) on chrysotile-cement products as dangerous as pure chrysotile fibres? Is the physicaland chemical composition of asbestos-cement dust different from pure asbestos dust?

Dr. de Klerk:

5.220 Risk from fibres depends on size, shape and durability (and their quantity). Asbestos cementcontains about 10-20 per cent asbestos, so that the dust concentration is going to be less than if thesheets were pure asbestos. However, cement does not form fibres, so that any airborne fibremeasurements made would only reflect the asbestos concentration in the air.

Dr. Henderson:

5.221 To answer the second part first, the physico-chemical composition of asbestos-cement dustdoes differ from pure asbestos dust, because the asbestos in asbestos-cement products is diluted by thecement (asbestos = 10-15 per cent by weight); this being so, one expects the asbestos fibres to bediluted by cement dust, in comparison to equivalent operations on pure asbestos materials.

5.222 To return to the first part of this question: the claim may be made that chrysotile fibresreleased from asbestos-cement products by high-speed cutting are altered physically or chemically,with a predominance of short-length fibres not implicated in carcinogenesis. For example, in Canada'sfirst oral submission, the following comments are made:

"The European Community has also advanced the thesis that the ban is necessary because France hasno control over trade persons or the 'handyman' who will cut into chrysotile-cement and, in so doing,free some of the chrysotile that was locked into it. Canada is puzzled by France's assertion that laRépublique Française is unable to regulate its handymen. In any event, there are three technicalreasons why France's concern is misplaced.

First, this thesis is based on the misconception that cutting high-density locked-in non-friable asbestos-containing materials releases substantial amounts of chrysotile. In fact, even if improper tools such ashigh-speed saws are used to cut chrysotile-cement, the dust released from such an operation containsonly a very small amount of pure, respirable-size chrysotile fibres, if any at all.

Second, science tells us that most of the chrysotile fibres released during high-speed cutting have beenchemically altered: the resulting entity is chemically and structurally different, and has a biological

WT/DS135/RPage 297

potential to induce harmful effects, which is different, and less than amphiboles. Similarly, the dustthat results from abrading chrysotile-containing resin or plastic reinforced products contains very smallamounts of chrysotile fibres. The same is also true of the dust that comes from wear and abrasion offriction materials: analysis of brake shoes shows that almost all of the chrysotile fraction of thefinished product is found to be transformed into a totally different, biologically-inactive material calledforsterite ...."

5.223 The first paragraph of the Canadian statement is dealt with in later discussion in this report(my answers to Question 5). For the second and third paragraphs of this statement, one can state thatin other situations, only a small fraction of airborne asbestos fibres are of respirable size: as oneexample, about 0.67 per cent of airborne asbestos fibres within indoor air of buildings were longerthan 5 µm in length. However, in the Japanese study reported by Kumagai et al. [4] on cuttingasbestos-cement pipes with a high-speed disc cutter, where airborne fibre concentrations within thehole used to gain access to the pipes averaged 92 fibres/ml (range 48-170 fibres/ml), the study dealtwith fibres longer than 5 µm (i.e. dimensions in the range for which carcinogenicity has beenreported). Please see also Table 11 in EHC 203 where Rödelsperger et al. recorded airborne fibreconcentrations of 4-5 f/ml and 5-10 f/ml for fibres longer than 5 µm, from blowing off and grindingbrake blocks, including truck brakes. (Please see also Table 11 and my answer to Question 2.)

5.224 Clearly, there is disagreement between the parties to the dispute and their respective expertsover the issue of whether chrysotile fibres released from high-density products are dangerous. For thereasons outlined above and in later discussion, it is my perception that at least a small proportion ofthe fibres has dimensions that are associated with carcinogenicity.

Dr. Infante:

5.225 As long as the interventions result in the release of chrysotile fibres, the exposures should beconsidered as dangerous as pure chrysotile fibres because respirable-sized asbestos fibres will bereleased. The study by Spurny (1989) indicates that the fibres released from weathered and corrodedchrysotile asbestos cement products have the same carcinogenic potency as "standard" chrysotilefibres. Although fibres released by weathering may be somewhat different from fibres released fromcutting or drilling asbestos cement products, the former fibres show a potency similar to that of purechrysotile fibres. Moreover, because of the potential for cleavage during interventions on asbestoscement products, the dust resulting from cutting, drilling, etc. on asbestos cement may actually containa greater portion of the asbestos fibres being thinner and more respirable than those that were initiallymixed into the cement during the manufacturing process. Therefore, the fibres released from cementduring interventions should be considered at least as dangerous as "pure chrysotile fibres." I can findno data to support an opinion that fibres released from interventions on chrysotile asbestos cementproducts would be less carcinogenic, or less dangerous. Further, asbestos related pathology hasresulted from such situations.

5.226 The asbestos cement dust would be somewhat different in physical and chemical compositionfrom pure chrysotile asbestos because the cement dust would contain respirable asbestos fibres,crystalline silica plus other substances added to the cement.

Dr. Musk:

5.227 It is my opinion that in general airborne fibres released from asbestos-cement products pose arisk. This may differ from other sources of chrysotile depending on the characteristics of the fibres.The area of fibre characteristics and their relationship to different sources is not within my area ofexpertise.

WT/DS135/RPage 298

1.(g) What is the risk to human health associated with demolition and removal of high-densitychrysotile products, such as chrysotile-cement products? Can you quantify this risk?

Dr. de Klerk:

5.228 See my answers to Questions 1(c) and (d).

Dr. Henderson:

5.229 I am not aware of any studies that have specifically focussed on either of these situations:therefore, no firm data are available, but one would expect the biohazards to be related to cumulativedoses of respirable fibres (i.e. airborne fibre concentrations and the frequency of exposure from thesetypes of work). This being so, one would expect the risks to be equivalent to other operations of likefrequency that generated similar airborne fibre levels (Tables 12 and 13).

Dr. Infante:

5.230 Exposure to high density chrysotile products through demolition carries with it the potentialrisk of lung cancer, asbestosis and mesothelioma. Testimony presented at the OSHA hearings relatedto its Final Asbestos Standard that was promulgated in 1994 indicated that removal of intact chrysotileasbestos "transite" panels that were held in place by screws can result in airborne fibre concentrationsthat exceeded 1 f/cc. In this situation, the exposed surfaces were wet prior to removal and theoperation was done within a negative pressure enclosure. Many transite panels used in interior wallconstruction consist of rough inner surfaces from which asbestos fibre is readily released into the air.Other testimony (OSHA, 1994) presented evidence that transite panels can be removed in a mannerthat results in exposure well below 0.1 f/cc when appropriate work practices are followed. Because ofconcern for the potential release of asbestos fibres into the air from such demolition, the OSHAstandard requires that a "competent person" supervise such activities, e.g., make an assessment anddetermine that the type of controls being used are appropriate for the removal situation and that therequired work practices are being followed. Therefore, the extent of the risk during demolition ofchrysotile cement products depends upon the compliance with mandated requirements. (See myanswer to Question 5(c) regarding compliance with procedures to reduce risk of disease from asbestosexposure.)

Dr. Musk:

5.231 In so much as demolition activities may result in airborne fibres there is a risk (as above).

1.(h) What is the risk to human health associated with high-density chrysotile wastes, such aschrysotile-cement waste? Can you quantify this risk?

Dr. de Klerk:

5.232 This depends on how the waste is treated and stored, depending of course on the chance ofany fibres becoming airborne and hence respirable. Otherwise, see my answers to Questions 1(c)and (d).

Dr. Henderson:

5.233 See my response to Question 1(g).

WT/DS135/RPage 299

Dr. Infante:

5.234 I have not researched this issue, but I am inclined to believe that there would not be muchpotential for fibre exposure from the handling of such waste unless a person at a waste site washauling asbestos cement and not aware of the product he/she was moving.

Dr. Musk:

5.235 The risk posed by waste products will also be dependent on the chances of fibres becomingairborne as above.

1.(i) Can high-density chrysotile products wastes, such as chrysotile-cement wastes, be dealt withso as to eliminate risks to human health?

Dr. de Klerk:

5.236 They can, by following approved methods for disposal which ensure that fibres are sealedfrom airborne release. There is of course the chance that subsequent work (for example, wasteremoval) may disturb the waste and release fibres.

Dr. Henderson:

5.237 In theory, YES — once the asbestos-cement or other high-density product has been removedfrom its in-place location (though few data are available on exposure levels produced by the actualremoval). For example, in Australia, imported chrysotile is delivered to production facilities in sealedplastic bags, so that the same procedure for bagging or encapsulation of high-density wastes shouldalso be applicable, and should prevent release of asbestos fibres once the encapsulation or baggingexercise is complete, unless the bags are ruptured for one reason or another.

5.238 According to NICNAS 99 (p 74), in Australia:

"Waste chrysotile, the polyethylene bags in which it is supplied, and chrysotile containing materialsfrom the manufacturing process, are disposed to landfill by licensed disposal contractors. As chrysotilefibres are unlikely to be mobile in the soil or water table, landfill is not inappropriate from a publichealth perspective."

Dr. Infante:

5.239 I have not researched this issue.

Dr. Musk:

5.240 These risks may be eliminated if the fibres could be successfully sealed so that they cannotbecome airborne.

Question 2:

What is the risk to human health associated with other current applications of chrysotileasbestos (in particular, friction materials and textiles)? In occupational circumstances? In non-occupational circumstances?

WT/DS135/RPage 300

Dr. de Klerk:

5.241 While the industries themselves may be well regulated, controlled and compliant withstandards, the major problem again could occur in "downstream" users: boilermakers, plumbers, brakemechanics etc. Fibres released from friction products have a higher proportion of shorter fibres thanthose from textiles, which release the highest proportion of longer fibres.

Dr. Henderson:

(i) Friction Products (e.g. brake linings)

5.242 Automotive mechanics and garage workers constitute a large population of workerspotentially exposed to chrysotile derived from brake linings. For example, brake blocks and liningsused in Australia have contained only Canadian chrysotile for many years, and the materials are eitherimported as pre-formed brake blocks and linings, or chrysotile is imported into Australia forsubsequent manufacture of these products. It has been estimated that this group of mechanicsamounts to at least 900,000 workers in the US, and the figure may be even higher if one adds in allthose who have ever worked in the automotive repair industry but then moved into otheremployments, and those who have retired.

5.243 For Australia, the number of persons employed as mechanics in 1991 amounted to 85,155(84,293 males); for 1996, the corresponding figures are 83,647 (82,827 males), out of a totalpopulation of 16,852,256 in 1991 (8,363,677 males); for 1996, the total population was 17,892,423(8,849,224 males). These figures for Australia include all mechanics, including automotive, brake andengine mechanics, together with supervisors and apprentices; the figures for 1996 also includemechanics' assistants (not included for the 1991 figures).21 Taking into account the fact that theAustralian population is less than 1/10th that of the United States population, these statistics appear tobe roughly comparable.

5.244 The literature contains anecdotal reports of malignant mesothelioma among automotive andbrake mechanics. However, the question that arises is whether these anecdotal reports are explicableas the chance occurrence of spontaneous or background mesotheliomas among a large population ofmechanics, or whether this group of workers has sustained other significant exposures to asbestos,including one or more amphiboles. In other words, the question is whether there is a general increasein the incidence of mesothelioma among automotive and brake mechanics with no other exposures toasbestos.

5.245 Brake repair workers are potentially exposed to asbestos during a number of procedures,which include removal of dust from bakes by air hoses, and a variety of other manipulations thatinclude bevelling, grinding and drilling. Clearance of dust from brakes by use of an air hose cancreate a cloud of visible dust, and airborne fibre concentrations of 2.0 to 29.4 f/ml have been recordedin the immediate vicinity [225, 226]. Please see also Table 11 in EHC 203 (pp 42-43).

5.246 In North America, chrysotile has been used almost exclusively in brake linings sincethe 1940s; chrysotile is also the type of asbestos used in brake linings in Europe (and also Australia).As I mentioned previously, commercial chrysotile (e.g. Canadian chrysotile) contains on averagesmall quantities of contaminant amphiboles in the form of tremolite (usually ≤ 1%).

5.247 However, the significance of this type of potential exposure among brake repair workers iscomplicated by a number of factors:

21Statistics supplied by the Australian Bureau of Statistics on 12 October 1999.

WT/DS135/RPage 301

• During moderate braking of automobiles, temperatures as high as ≥ 500oC can bereached within the brakes, and at this temperature a proportion of the chrysotileundergoes dehydroxylation and recrystallization to form the mineral forsterite, whichis not implicated in mesothelioma induction.

"Heating of chrysotile at 700°C for an hour converts it to an amorphous, anhydrousmagnesium silicate material ... Intensive dry grinding also destroys the structure ofchrysotile. Analysis of wear debris from brake linings made with asbestos has shownthat virtually all of the chrysotile fibre is converted to amorphous material, inassociation with the mineral forsterite (a recrystallization product). The conversion isexplained by localized temperatures above 1000°C at the point of contact between thebrake lining and the drum" ... . [EHC 203, p 14].

• Most chrysotile fibres released from brakes comprise short-length fibres < 0.4 µm inlength (> 80 per cent of all chrysotile fibres from brakes). However, some fibres > 5µm in length and even > 10 µm in length appear to survive (e.g. please see Table 11in EHC 203). The short-length fibres appear to have only questionable or limitedcarcinogenicity, and this property is thought to reside primarily in fibres > 5 µm inlength. In addition, limited fibre burden studies on brake repair workers have showna low pulmonary asbestos content.

"The fibres found in the brake wear debris are predominantly (99%) less than 0.4 µmin length … Rödelsperger et al., (1986) found less than 1% of fibres longer than5 µm." [EHC 203, p 14]

• One also needs to remember that assessment of mesothelioma risk among brake repairworkers can be confounded by other occupational exposures to asbestos [227].

5.248 In a review of changing risk groups for malignant mesothelioma, Huncharek [228] gives thefollow account for brake mechanics:

"A major problem with epidemiologic studies of this workforce is the difficulty in tracing a large, non-unionized group of workers. Estimation of disease risk has been impeded by lack of quantitative dataon exposure levels among individuals with long-term exposure. ...

In 1976, investigators at the Mount Sinai School of Medicine studied asbestos exposure among brakerepair workers in New York city. Both clinical examinations and fibre counts produced by variousoperations in brake maintenance workers were analyzed. Samples taken at a distance of 3-5 feet frombrake drums during periods of blowing dust showed fibre concentrations of 6.6 to 29.4 fibres/ml, with amean of 15.9 fibres/ml. In addition, ten samples of brake drum dust were analyzed by phase contrast,optical microscopic examination and transmission electron microscopic examination to determine thepercentage of short fibres (i.e., 25-500 angstroms x 760 to 3750 angstroms22). Eighty-three percent ofall chrysotile fibres were in this category, and almost 20% of the total mass of 10 samples waschrysotile (determined by electron diffraction). 'Throughout the examination by electron microscopy,attention was given to the morphology of the fibres. A majority of fibres showed little alteration in thetypical chrysotile fibre'.

In an additional report from Mount Sinai, Rohl et al. analyzed residual dust recovered from brakelinings and made direct measurements of the free asbestos fiber content of 'workroom air' in areas inwhich brake lining maintenance and brake shoe installation occurred. Airborne asbestos dustconcentrations were similar to those cited by Lorimer et al. (i.e., mean airborne fiber concentrationsduring compressed air blowing of brake drums ranged from 2.6 fibers/ml at a distance of 10-20 feet to16.0 fibers/ml at 3-5 feet). Samples of brake drum dust showed that the proportion of chrysotile in thismaterial averaged 3% to 6% (both as free fibre and as particulates in pulverized binder).

2210,000 Ångstroms = 1.0 µm.

WT/DS135/RPage 302

Regarding the health effects resulting from the exposures described, Langer and McCaughey published... a case of mesothelioma in a brake repair worker ... a 55-year-old man who had worked in the usedcar, tire and car repair business since the age of 19 years. He reported routinely servicing automobiles,including the replacement of brake linings. No other source of asbestos exposure was found.

Analysis of lung tissue showed the presence of chrysotile fibres (no amphibole was found) confirmedby electron diffraction techniques. Ten percent of the fibers found were longer than 10 microns.23 Theauthors point out that 'Controversy over the potential of chrysotile to cause mesothelioma has continueddespite evidence from asbestos textile fabricators, thought to have used only chrysotile, from workersmaking brake pads, from chrysotile miners and millers, and from animal studies'. They also state 'Therisk of malignant asbestos disease among these workers seems to be low but mortality data have yet tobe thoroughly evaluated.'

The most recent report of mesothelioma in a brake mechanic reviews a pleural mesothelioma occurringin a 47-year-old male automobile mechanic whose only known exposure to asbestos was from clutchand brake repair work during an 11-year period. ...

Another case of mesothelioma in a brake repair worker was recently published. In this report, a56-year-old male elevator mechanic ... reported working as an elevator mechanic for 30 years. Hereported exposure primarily from elevator brake linings that he routinely cut, fitted, and removedduring elevator installation and maintenance.

Several recent studies from Scandinavia on this topic also deserve mention. Hansen, from the Instituteof Community Medicine in Denmark, completed a historical cohort study examining the mortality ofcar mechanics from ischaemic heart disease and malignant neoplasms. The study cohort was identifiedusing records of a nation-wide census carried out in Denmark in November 1970. Comparison wasmade with another cohort of skilled male workers who were not exposed to asbestos or 'petrochemicalsubstances'. Of 583 observed deaths, one case of pleural mesothelioma was found.

Likewise, Jarvholm and Brisman, in a 1998 report, used the Swedish death register and the census of1960 to study the occurrence of asbestos-associated tumors in car mechanics. One hundred and eighty-seven deaths attributable to cancer were observed, whereas 154 were expected. Thirty-nine werecaused by lung cancer, whereas only 23 were expected. Again, one death from pleural mesotheliomawas found. ...

It has been estimated that 20,000 deaths from asbestos-related cancer will occur during the next 40years among automotive maintenance workers in the United States. With the many difficulties facedby epidemiologists studying this workforce, it is unclear how accurate this estimate will prove to be.Clearly, what is needed is better information on duration and intensity of exposure to respirableasbestos fibers in this occupational group. Additional study is needed to accurately determine theincidence of mesothelioma among members of this workforce." [pp 2704-2705].

5.249 The situation is further complicated by other reports on garage mechanics and workersinvolved in friction products manufacture [229, 230]. These studies have been reviewed briefly byWong [231]: three found no increase in the RR for mesothelioma among garage mechanics (RR =0.9, 0.65 and 1.0 respectively).

5.250 An analysis of > 13,000 workers at a UK friction products factory showed no detectableexcess mortality due to lung cancer or other cancers; 13 mesotheliomas were found but 11 had knownexposure to crocidolite [229, 232].

23This relatively high proportion (10 per cent) in comparison to the smaller fraction of airborne fibres of

the same size is presumably explicable by preferential clearance of short fibres from lung tissue, with aproportional increase of long fibres over time.

WT/DS135/RPage 303

5.251 McDonald et al. [230] identified excess mortality from lung cancer among friction productsworkers, but there were no mesotheliomas:

"A study by McDonald et al. (1984) investigated mortality due to lung cancer, mesothelioma andasbestosis in three US factories manufacturing friction products and packings. The cohort comprised3641 men employed between 1938-1958. During the 1930s exposures for most processes were 1-5mpcf (millions of particles per cubic foot) and > 10 mpcf during dry mould mixing. By the 1960s mostexposures were < 0.5 mpcf. A significant excess of deaths (reference was to mortality rates forConnecticut) due to respiratory cancer was observed however this was not related to duration ofemployment. No cases of mesothelioma were reported. There was limited evidence of an increase inrisk of lung cancer with increasing exposure. However the SMR for lung cancer was noted in workerswith less than one year of service.

A study by Finkelstein, (1989) investigated mortality rates among 1657 employees at two Ontariofactories manufacturing chrysotile friction materials. The study population consisted of workersemployed for at least 12 months after 1 January 1950. The study showed a significant increase inmortality from laryngeal cancer and lung cancer. No increase in mortality was noted fromgastrointestinal cancer or from non-malignant respiratory disease. One or two deaths may have beendue to pleural mesothelioma. Case-control analysis demonstrated a lack of association between the riskof death from laryngeal or lung cancer and the duration of employment or employment in departmentswhere chrysotile had been used. The author also noted that cigarette smoking is a risk factor forlaryngeal cancer and lung cancer, and therefore, increased risk may be in part attributable to differencesin smoking habits." [NICNAS 99, p 65].

5.252 Similarly, Woitowitz and Rödelsperger [227, 233] found that:

"There is no evidence that car mechanics are exposed to an increased risk of mesothelioma even if theydo brake repairs, but asbestos exposure in other employment is an important confounding factor, so thatif there is a mesothelioma risk for car mechanics but it was small, it would not be detectable."

5.253 Nonetheless, the 1999 Report for the Australian Mesothelioma Register24 (AMR 99) records58 mesotheliomas among brake mechanics with no other exposures to asbestos, during thealmost 13-year period between 01 January 1986 and 31 October 1999 (total cases with a stated historyof asbestos exposure = 2585). Mechanics who frequently or consistently work on brake linings andbrake blocks represent only a sub-fraction of the total workforce of mechanics in Australia. If onetakes the 1996 census figure of 82,827 for male mechanic25, this amounts to 58 mesotheliomasin 1,062,946 person-years (≡ 54.6 mesotheliomas per million person-years). If one rounds off theworkforce to 100,000 male mechanics, the figure becomes 45 mesotheliomas per million person-years. If one then doubles the workforce population to take into account retirees and other workerswho moved on to other occupations (although a figure of 200,000 is almost certainly an overestimatebecause it would include all mechanics, whereas brake mechanics constitute a smaller sub-class), themesothelioma rate becomes 22.6 per million person-years — well under the rateof 337 mesotheliomas per million person-years for the Quebec chrysotile miners and millers but stillsubstantially above the upper limit of the estimated background rate of 1-2 mesotheliomas per millionperson-years (about 10-fold). One might suspect that mesotheliomas in brake mechanics will clusterin those involved in the grinding, bevelling and other operations on new brake blocks and brakelinings (i.e. brake materials unaltered by heat).

5.254 Using an earlier set of data for Australia, NICNAS 99 came to a similar conclusion:

24The Register is a compilation of all and unselected mesotheliomas throughout Australia.25This over-estimates the number of brake mechanics, because the figure includes all automotive

mechanics, engine mechanics, apprentices, and supervisors: Australian Bureau of Statistics, 12 October 1999.

WT/DS135/RPage 304

"Out of 2119 mesothelioma cases registered (with a response to history) for the period 1986-1995,46 cases were listed for the category 'brake lining - manufacture/repair', 40 of which were recorded incar mechanics, of which 37 were exposed to asbestos in this occupation only ... Overall the numbersindicate a slight increase of around 1-2 cases per year, which is roughly proportional to the growth rateof all mesothelioma cases in Australia" ... [p 66].

5.255 It is apparent that these considerations apply to occupational circumstances.

5.256 Evidence indicates that the general population is exposed to only very low levels of asbestosderived from the braking of passing automobiles, and that most of these fibres represent short-lengthfibres and heat-altered chrysotile. NICNAS 99 has this to say on the subject:

"It is claimed that the amount of asbestos found in the dust arising from braking is rarely more than 1%of the wear product (Asbestos Information Committee, 1975). It is not known what quantity ofchrysotile is imported in brake linings and other friction materials, but ABS [Australian Bureau ofStatistics] data indicates in excess of 750,000 articles (brake linings, pads and clutch facings) beingimported in 1997 containing asbestos and therefore possibly containing chrysotile. Assuming each unitweighs 200 g and contains 50% chrysotile, this equates to around 150 tonnes of chrysotile per annum.Assuming a further 1000 tonnes of chrysotile present in friction products manufactured in Australia, itis estimated that (assuming a worst case scenario of 1% release per annum, i.e. all products arecompletely worn in one year), around 11.5 tonnes of chrysotile will be released per annum countrywideor 32 kg per day spread all around the country. It is acknowledged that this figure may be anoverestimate, as studies have shown that some of the chrysotile is degraded to magnesium silicates andforsterite ... In addition, some of the debris will be retained in the brake system and removed anddisposed of under controlled conditions." [p 78].

(ii) Exposure from chrysotile textile materials

5.257 Table 14 represents a reproduction of Table 7 in EHC 203 for exposure estimates in the SouthCarolina chrysotile textile plant (1930-1975) before and after controls on exposure levels. As can beseen from this Table, the application of controls on exposure levels produced a significant reduction ofexposure, and currently available control technology allows even lower levels to be attained(EHC 203).

5.258 EHC 203 refers to a study in Japan which reported a geometric mean concentrationof 0.1-0.2 f/ml in the period 1984-1986, for asbestos spinning. From published studies, it seems clearthat asbestos textile workers are at greater risk for asbestosis (historically) and lung cancer thanmesothelioma. EHC 203 also gives the following comments:

TABLE 14: EXPOSURE ESTIMATES IN A CHRYSOTILE TEXTILE PLANT (1930-1975)(ESTIMATED MEAN EXPOSURE TO FIBRES LONGER THAN 5 µM IN F/ML)

Operation Without controls With controls

Fibre preparation 26.2-78.0 5.8-17.2

Carding 10.8-22.1 4.3-9.0

Spinning 4.8-8.2 4.8-6.7

Twisting 24.6-376.0 5.4-7.9

Winding 4.1-20.9 4.1-8.4

Weaving 5.3-30.6 1.4-8.2

From Dement et al. (1983)

WT/DS135/RPage 305

"Studies that correlate disease prevalence or symptoms with cumulative exposure can under-estimatedisease risk due to progression of disease after employment ceases. Although workers were exposed toboth chrysotile and crocidolite (the latter being approximate 5% of all asbestos used), resultsfor 379 men employed at least 10 years in the Rochdale asbestos textile plant are informative ...Exposure estimated from work histories range from an average of 2.9 to 14.5 f/ml. Overall, smallopacities (> 1/0) were reported in 88/379 (23%) of chest radiographs, with evidence of a gradientseriously confounded by date of first employment and transfer of subjects with suspected asbestosis toless dusty conditions. On the basis of data on incidence, the authors drew conclusions on exposure-response between cumulative exposure and prevalence or incidence of crepitations, possible asbestosisand certified asbestosis - all three depending on clinical opinion and judgement. The authors state thatpossible asbestosis occurs in no more than 1% of men after 40 years of exposure to concentrationsbetween 0.3 and 1.1 f/ml" [EHC 203, p 105].

5.259 On p. 114, EHC 203 goes on to discuss other consequences of exposures sustained duringtextile manufacture:

"The health of employees has been studied in any detail in only three asbestos textile plants. Thesecomprise a factory at Rochdale, England, originally studied by Doll (1955) and more recently by Petoet al. (1985), another located in Mannheim, Pennsylvania, USA, studied by McDonald et al. (1983b)and a plant in Charleston, South Carolina, USA. Only the study in South Carolina is consideredprimarily relevant for assessment of the health effects of chrysotile. Although the SMRs for lungcancer in these plants were broadly equivalent, the rates of mesothelioma varied considerably, whichmay reflect the greater proportions of amphiboles in the Mannheim and Rochdale cohorts.

The textile workers in South Carolina plant have been studied in two separate but overlapping cohorts... . The only amphibole used in this plant was approximately one tonne of imported crocidolite fromthe early 1950s until 1972, plus a very small quantity of amosite for experimental purposes briefly inthe late 1950s. The crocidolite yarn was processed at a single location only, so Charleston can beconsidered an almost pure chrysotile operation. Exposure levels for workers at this plant wereestimated by Dement et al. (1983a) using nearly 6000 exposure measurements covering the period1930-1975 and taking into account changes in plant processes and engineering controls (Table 7). Theconversion of past exposures measured in mpcm (mpcf) to f/ml was based on both paired sample data(100 pairs) and concurrent samples (986 samples) by these two methods collected in plant operationsduring 1968-1971.

The most recent update of the Charleston study by Dement et al. (1994) demonstrated an overall lungcancer SMR of 1.97 (126 observed) and an overall SMR for non-malignant respiratory diseases ... of3.11 (69 observed). The data for white males, for which data were more complete, demonstrated anoverall lung cancer SMR of 2.34 for those achieving at least 15 years of latency. The risk of lungcancer was found to increase rapidly in relation to cumulative exposure. Data for the entire cohortdemonstrated an increase in the lung cancer risk of 2-3% for each fibre/ml-year of cumulativechrysotile exposure. Two mesotheliomas were observed among this cohort and an additionalmesothelioma was identified among plant workers, occurred after the study follow-up period. Analysesof an overlapping cohort from the same factory ... provided similar results.

... the regression line slopes for relative risks of lung cancer in relation to accumulated exposure in theCharleston plant are all some 30 times steeper than those observed in chrysotile mining and cementproduct manufacture."

5.260 From the foregoing discussion, it is plain that these risks apply to occupational circumstances,and not to non-occupational situations. It is my perception that there is debate among experts over thecarcinogenicity of chrysotile released from or associated with friction products and textilesrespectively.

WT/DS135/RPage 306

Dr. Infante:

5.261 Exposure to chrysotile asbestos through the manufacturing and downstream manipulation offriction products and textiles carries with it the risks associated with exposure to asbestos, mostnotably, lung cancer, asbestosis and mesothelioma. This is mostly an occupational issue except forconsumers, who do their own brake replacement which would put them at risk of developing thesediseases.

5.262 Epidemiological studies of workers involved in the manufacturing of friction productsdemonstrate an elevated risk of lung cancer (McDonald et al. 1994). Other investigators have notobserved an excess of lung cancer in the manufacturing of chrysotile and cricidolite containingfriction products, but have identified cases of mesothelioma related to both types of asbestos used toproduce these products (Berry and Newhouse, 1983). Cases of mesothelioma also have been reportedamong car mechanics who serviced brakes containing chrysotile fibres only and were exposed tolevels estimated to be below 1 f/cc-year cumulative exposure (Woitowitz and Rodelsperger, 1991).Epidemiological study results indicate a high risk of disease related to chrysotile textiles. See myresponses to Questions 4(a)-4(c).

Dr. Musk:

5.263 It is my opinion that there is a risk of disease from the release of chrysotile fibres from frictionmaterials (as in brakes and clutches) or textiles (as in asbestos blankets and suits). In general the riskwill be dependent on the degree of exposure (as above) and is therefore likely to be greater in peoplewith occupational rather than non-occupational exposure. Fibres released from friction materials maybe shorter than from other sources. These fibres may clear more rapidly from the lungs and possiblybe related to lower risks but I am not aware of any direct data on this.

Question 3:

The parties disagree as to the relative pathogenicity of amphibole and chrysotile asbestos.Canada argues that, due to chemical and physical differences, a crucial distinction is to be madebetween amphibole and chrysotile asbestos: the latter is less pathogenic than the former. TheEuropean Communities, on the other hand, argues that chrysotile is as dangerous as amphiboles.In answering the four sub-questions below, please specify to what extent your opinion is based onepidemiological data, in vivo or in vitro evidence.

3.(a) For the purpose of assessing the risk to human health arising from exposure to asbestosfibres, should a distinction be made between chrysotile and amphibole asbestos?

Dr. de Klerk:

5.264 In terms of risks to human health, epidemiological evidence is clear that, for a given quantity(intensity and duration) of exposure, chrysotile imparts less risk than amphibole fibres.

Dr. Henderson:

5.265 YES — a clear distinction should be made between chrysotile and the amphibole forms ofasbestos. On a fibre-for-fibre basis, amphiboles such as crocidolite and amosite are substantially morecarcinogenic for the mesothelium than chrysotile. This potency differential is to some extentconfounded by the far greater usage of chrysotile both now and historically (> 95 per cent of worldasbestos production). In addition, it is worth reiterating that Canadian chrysotile on average containstrace quantities of fibrous tremolite (an amphibole).

WT/DS135/RPage 307

5.266 It is my perception that there is broad agreement among experts that the amphiboles are morepotent carcinogens for the mesothelium than chrysotile.

Dr. Infante:

5.267 For purposes of assessing disease risk from exposure to asbestos fibres, I see no basis formaking a distinction between chrysotile and amphibole asbestos. Several high qualityepidemiological studies of workers exposed to chrysotile asbestos demonstrate an elevated risk ofdeath from lung cancer, asbestosis and mesothelioma. The risk of death from lung cancer andasbestosis related to chrysotile exposure appears to be similar to that from exposure to other forms ofasbestos. Although epidemiological study suggests that the risk of death from mesothelioma fromchrysotile exposure may be less than the mesothelioma risk from amphibole asbestos, it is somewhatdifficult to make this comparison. Many of the studies of workers exposed to amphibole asbestos goback further in time and less information is available on the quantitative aspects of exposure. Thus,the role of error in exposure estimation on the perceived difference is difficult to determine. On theother hand, some experimental inhalation studies demonstrate that chrysotile asbestos may be morepotent than other forms of asbestos in the induction of mesothelioma (and lung cancer) in relation tothe amount of dust deposited in the lungs (Wagner et al., 1974). In any event, the attributable risk ofrespiratory diseases from exposure to asbestos is going to be heavily weighted by lung cancer andasbestosis. So, even if exposure to chrysotile asbestos would result in a slightly lower relative risk ofdeath from mesothelioma, the overall risk of the asbestos related diseases combined, i.e., lung cancer,asbestosis, mesothelioma, decrements in pulmonary function, will not be perceptively different forchrysotile as compared to the amphiboles. Thus, a distinction should not be made between chrysotileasbestos and amphibole asbestos. In my opinion, exposure to all forms of asbestos results in asignificant disease burden to society.

5.268 I believe that there is a need to distinguish chrysotile asbestos from amphiboles based on theepidemiological data at least.

3.(b) What are the key properties which cause pathogenicity of, respectively, amphiboles andchrysotile fibres for (i) asbestosis, (ii) lung cancer, (iii) mesothelioma, and (iv) other asbestos-related pathologies?

Dr. de Klerk:

5.269 The properties are the same: size, shape and durability in the lung, that is fibres have to be ofcertain size and shape to be deposited in the lungs and have to stay there long enough to produce aresponse. Since most of the body's responses to fibres appear to be stochastic in nature, the additionalfeatures are of course the intensity and duration of exposure as outlined above. All the types ofasbestos differ according to these properties, the main difference with chrysotile is that it is lessdurable in lung tissue than amphibole fibres: it is more soluble and the fibres tend to break morereadily into smaller fibrils, it also tends to be more curly rather than dead straight in shape.

Dr. Henderson:

5.270 As discussed already, the pathogenicity of asbestos appears to reside in the physical propertiesand biopersistence of the fibres, summarized as the 3 Ds — namely dose, fibre dimensions anddurability (bio-persistence). All commercial asbestos has the capacity to induce asbestosis, lungcancer, mesothelioma and other pleural abnormalities (e.g., parietal pleural fibrous plaques, benignasbestos pleuritis with effusion, and diffuse pleural fibrosis). Chrysotile and the amphiboles havedifferent potencies in generating these disorders: for example, the amphibole varieties of asbestosappear to be substantially more pathogenic than chrysotile for the induction of asbestosis andmesothelioma, whereas the differential is not sustained for the induction of lung cancer, for which

WT/DS135/RPage 308

chrysotile is associated with one of the lowest lung cancer rates (in Quebec chrysotile miners andmillers) and the highest rate of lung cancer (South Carolina asbestos textile workers, who usedCanadian chrysotile).

5.271 Asbestosis: There is good evidence that asbestosis is a dose-dependent disorder with athreshold effect. There is widespread agreement that asbestosis in general is a consequence of highintensity exposure (or lower intensity but more prolonged exposure) than mesothelioma not associatedwith asbestosis, so that the concentration of asbestos bodies and uncoated asbestos fibres within thelung tissue of asbestotics is considerably higher than the concentrations encountered in mesotheliomapatients without asbestosis and in individuals with parietal pleural plaques. In addition, some studieshave shown that the severity of asbestosis and its liability to progression are related to the asbestosbody and fibre concentration in lung tissue. In the past, asbestosis as a consequence of high-doseexposure was a progressive disorder leading to progressive respiratory insufficiency and death,whereas many cases of asbestosis encountered during the 1980s and 1990s represent milder and staticforms of the disease.

5.272 Churg [234] points out that:

" ... a considerably greater burden of chrysotile (with its accompanying tremolite) than of amosite orcrocidolite is required to produce any particular diseases. ... For example, the mean burden ofchrysotile plus tremolite in the lungs of [Quebec chrysotile] miners and millers with asbestosis is 17times the amosite burden in the lungs of shipyard workers with asbestosis." [p 294].

5.273 To some extent, this differential in potency may reflect fibre dimensions and the generation ofoxidants, but a more likely explanation is the faster clearance of chrysotile from lung tissue than anyof the amphiboles. However, Churg's comments seem to apply to Quebec chrysotile miners andmillers in particular; in contrast, the study reported by Green et al. [191], which reported histologicalasbestosis at relatively low cumulative exposures was carried out on South Carolina asbestos textileworkers who also worked with Canadian chrysotile (please see following discussion).

5.274 There is also evidence that long fibres are implicated in the development of asbestosis, butthis may reflect in part the anatomical distribution of long versus short fibres in lung tissue. Forexample, some studies have shown that short fibres in the vicinity of 1 µm in length have a biologicalpotency similar to that of the longer fibres for initiation of the inflammatory changes implicated in thedevelopment of asbestosis, but they do not penetrate the walls of bronchi or bronchioles to the sameextent as the longer fibres, to reach the alveolar interstitium. Another study related the developmentof asbestosis to the total surface area of deposited fibres, rather than to particular lengths of fibres.

5.275 The effect of dose: There is good evidence that the inhaled dose of asbestos affects: (i) thedevelopment of the disease itself; (ii) the latency period between exposure and the onset of thedisease; and (iii) the severity and progression of disease.

5.276 Fibre burden studies on human lung tissues show that patients with asbestosis in general havehigher tissue burdens than patients with asbestos-related diseases other than asbestosis. Accordingly,Mossman and Churg [202] state that:

" ... asbestosis is clearly a fiber-dose driven disease but, nonetheless, only a fraction of any cohortexposed to a fibrogenic dose of asbestos develops asbestosis. It has been proposed that person-to-person variations in either fiber deposition or fiber clearance may account for this phenomenon."[p 1671].

5.277 The Ontario Royal Commission [235] noted that asbestos exposures < 25 fibre-years wouldbe unlikely to produce clinical asbestosis (about ≤ 1 per cent of individuals exposed at this level may

WT/DS135/RPage 309

develop clinical or radiological asbestosis), whereas Browne [236] considered that the minimum doserequired to produce clinical asbestosis was in the range of 25-100 fibre-years.

"Chest X-ray changes among textile and friction product workers in China were reported by Huang(1990). A total of 824 workers employed for at least 3 years in a chrysotile products factory from thestart-up of the factory in 1958 until 1980, with follow-through to September 1982, were studied. ChestX-ray changes compatible with asbestosis were assessed using the Chinese standard system forinterpretation of X-rays. Cases were defined as Grade I asbestosis (approximately equivalent toILO ≥ 1/1). Overall, 277 workers were diagnosed with asbestosis during the follow-up period,corresponding to a period prevalence of 31%. Exposure-response analysis, based on gravimetric dataconverted to fibre counts, predicted a 1% prevalence of Grade I asbestosis at a cumulative exposureof 22 f/ml-years." [EHC 203, p 106]

5.278 In an autopsy study of South Carolina asbestos textile workers who used Canadian chrysotile— the same group studied by Dement et al. [171, 172, 237-241] and McDonald et al. [161, 242] —Green et al. [191] showed that histological asbestosis was usually present at ≥ 20 fibre-years ofexposure, and a few cases were observed at 10-20 fibre-years. Thimpont and de Vuyst [243] reportedthat around 50 per cent of specimens of lung tissue removed because of lung cancer showed low-gradeairway and interstitial fibrosis with asbestos bodies, when the asbestos body concentration was ≥ 5000per gram dry lung tissue.

5.279 These different threshold doses are not inconsistent with each other, because they deal withidentification of asbestosis by different modalities (i.e. clinical/radiological asbestosis versushistological asbestosis). In this respect, histological examination is generally considered to representthe most sensitive and specific technique for the diagnosis of asbestosis, followed in descending orderby high-resolution CT scanning, conventional CT scans and chest X-rays (which will fail to detectasbestosis in about 20 per cent of cases, especially low-grade asbestosis). In other words, early(grade I) asbestosis may be undetectable by clinical investigations.

5.280 Latency interval: there is also evidence that the latency period between first exposure toasbestos and the subsequent diagnosis of asbestosis is roughly inversely proportional to the exposurelevel, so that short latencies are encountered with high exposures (e.g. the Wittenoom cohort).

5.281 Mossman and Churg [202] also state that:

"Fibre burden studies also indicate that there is a correlation between pathologic severity of asbestosisand increasing burden of asbestos bodies (which are largely markers of amphibole exposure) oruncoated amosite and crocidolite fibres" ... . [p 1670].

5.282 There are two additional points that are worth emphasis:

• There appears to be considerable variation among individuals in their propensity todevelop asbestosis (or variation in the latency intervals for equivalent exposures). Forexample, I have seen cases where there was a high content of amphibole asbestos inlung tissue (up to 100 million fibres per gram dry lung tissue or more), in the absenceof histological asbestosis — at the time when the fibre burden analysis was carriedout — whereas other cases had clinical asbestosis with much lower tissue fibreburdens. Of course, one caveat about this situation concerns latency, whereas anotherrelates to the time when the fibre burden analysis was carried out, which is differentfrom the time when the disease was developing.

• In addition to high fibre burdens, the severity and progression of asbestosis can be

WT/DS135/RPage 310

influenced by other factors, such as cigarette smoke (though tobacco smoke by itselfcannot cause asbestosis, of course).

5.283 Mesothelioma: Please see discussion in Section C.1.(f) to(h).

5.284 Lung cancer: Please see discussion in Section C.1.(i).

5.285 Other asbestos-related pathology: The effect of dose is less clear for the induction of parietalpleural fibrous plaques than for other asbestos-related disorders. There is evidence that the frequencyand extent of plaques are dose-related, so that plaques tend to be more extensive with higherexposures. However, plaques can also follow trivial exposures to asbestos and asbestos-like minerals,so that the frequency of asbestos-related pleural plaques appears to correlate more closely withduration since exposure than the level of exposure. Plaques are known to be endemic in Finland,apparently as a consequence of very low-level exposure to asbestos-like fibres in the generalenvironment; on the other hand, in societies where plaques are not endemic — e.g. North America,Western Europe and Australia — about 80-90 per cent of radiologically well-defined plaques are aconsequence of occupational exposure to asbestos. In such societies, pleural plaques also represent auseful tissue marker for prior asbestos exposure. At the same time, it is also worth emphasising thatparietal pleural plaques by themselves do not predispose to any other asbestos-related disorder, theliability being related to inhaled dose and fibre types.

Dr. Infante:

5.286 The key properties of the pathogenicity of amphiboles and chrysotile fibres, although notknown for certain, are thought to be related to the physical characteristics of the fibres, namely thediameter, length, aspect ratio of length to diameter, surface area and perhaps surface charge of thefibres. Toxicological data suggest that long thin asbestos fibres may be relatively more potent thanother asbestos fibres in their ability to induce asbestos related diseases. There is also experimentalevidence, however, that shorter fibres can produce asbestos related diseases, though not in as great amagnitude. Any manipulation of these fibres that results in their diameter becoming thinner mayprovide a relatively greater contribution to the associated pathology and a greater toxic response thanif the fibres were not manipulated in any way. The issue of solubility has been raised in terms ofasbestos fibres and their capability to cause disease. The role of solubility with asbestos fibres anddisease potential, however, is not so clear because chrysotile asbestos seems to have the same overallpotency as other forms of asbestos, yet chrysotile fibres appear to be relatively more soluble thanamphibole fibres.

Dr. Musk:

5.287 It is my opinion that the key characteristics determining the pathogenicity of asbestos forasbestosis, lung cancer, mesothelioma and other diseases are determined by the physical and chemicalcharacteristics of the asbestos. The physical characteristics include length, diameter and"straightness" of the fibres. The chemical properties determine the durability of the fibres.

3.(c) What is the respective capacity of amphiboles and chrysotile to induce (i) asbestosis,(ii) lung cancer, (iii) mesothelioma, and (iv)other asbestos-related pathologies?

Dr. de Klerk:

5.288 Comparisons between chrysotile and the amphiboles in their capacity to producemesothelioma and lung cancer have been extensive, for the other diseases, much less so. There issome in vitro and in vivo evidence that amphiboles, particularly crocidolite, are more fibrogenic thanchrysotile, but there is no clear epidemiological evidence on this. Pleural plaques appear to be more

WT/DS135/RPage 311

common among anthophyllite workers than others while crocidolite workers have more diffuse pleuralthickening, and benign asbestos pleurisy also seems to be more common after crocidolite exposure.Historically, asbestosis occurred commonly after heavy exposure to all types of asbestos. Formesothelioma, it is thought that for a given cumulative exposure, chrysotile is between one tenth andone hundredth as potent as crocidolite. There is some controversy over the relative capacity ofamosite and crocidolite, but amosite seems to carry about one tenth of the risk of crocidolite. For lungcancer, amosite and crocidolite seem to have similar capacities, with chrysotile weighing in aroundone tenth to one fiftieth of this.

Dr. Henderson:

5.289 See my response to Question 3(b).

Dr. Infante:

5.290 The quantitative risk of dying from lung cancer as a result of exposure to chrysotile asbestosis at least as great as that from exposure to other forms of asbestos. Quantitative risk assessmentsbased on several epidemiological studies indicate a very high risk of lung cancer (potency) amongworkers exposed to chrysotile asbestos. The cohort study by Dement et al., first published in 1983and updated in 1994 contains one of the best estimates of worker exposure to chrysotile accompaniedby the estimates of relative risk (RR) for lung cancer and asbestosis. In this study, the investigatorsdetermined conversion ratios from the available industrial hygiene data by evaluating sampling resultsfrom surveys that used the impinger method to measure dust in million particles per cubic foot(MPPCF), and membrane filter samples that allowed for the counting of fibre concentrations. Forearly exposure periods, the investigators converted one MPPCF to the equivalent of 3 f/cc > 5 um inlength for all areas except for preparation where a conversion factor of 8 fibres for every MPPCF wasused. In my opinion, this study is the strongest in its methodology for estimating asbestos exposureamong cohort members. Quantitative risk assessment based on data for the entire cohort estimates anincrease in the RR of lung cancer ranging from 2-3 per cent for each fibre/cc-year of cumulativechrysotile exposure. To my knowledge, this is the highest estimated risk of lung cancer amongasbestos exposed workers that has been corroborated by other investigators, i.e., the same populationwas also studied by McDonald et al., (1983) and the results are remarkably similar. Studies of lungcancer risk among chrysotile asbestos textile workers from two additional occupational cohorts byMcDonald et al., (1982) and Peto et al., (1985) also provide similar results. Using data from theMcDonald et al., (1983) study and a conversion factor of 6 fibres per MPPCF, Peto et al., (1985)estimated an increase in the RR of lung cancer among chrysotile textile workers to be 1.25 per centper fibre/cc-year of exposure. From their own study of Rochdale textile workers, Peto et al., (1985)estimated the excess lung cancer risk to range from 0.5-1.5 per cent per f/cc-year of cumulativeexposure depending upon whether the estimate was based on the entire cohort or on those employed in1951 or later, respectively. These estimates are of surprisingly similar magnitude given that they arefrom epidemiological studies that incorporated retrospective exposure estimation that had to rely uponthe conversion of measurements from particles to fibres per cc. Thus, three separate populations ofchrysotile asbestos textile workers demonstrate remarkably similar elevated risks of lung cancer andtherefore add to the confidence of the estimates of excess lung cancer risk per unit of fibre exposurethat these studies demonstrate.

5.291 Studies of workers exposed to chrysotile in several industries demonstrate a significantlyelevated risk of lung cancer. Studies of workers exposed to a combination of chrysotile andcricidolite, or to chrysotile only in cement production (Hughes et al., 1987) indicate a virtuallyidentical excess risk of mortality from lung cancer. In a study of workers exposed to crocidolite inmining, de Klerk et al., (1989) estimate an elevated relative risk of lung cancer of 1 per cent per f/cc-year of exposure. Essentially, the excess relative risk for lung cancer indicated in all of these studiesis close to about 1 per cent per fibre per cc-year of exposure (Stayner et al., 1997). The risk of lung

WT/DS135/RPage 312

cancer from chrysotile asbestos exposure is at least as great as the risk of lung cancer associated withamphibole asbestos exposure.

5.292 Analyses based on the study by McDonald et al. (1993) indicates a much lower dose responsefor lung cancer in relation to chrysotile asbestos in mining and milling as compared to the riskestimates from other studies, particularly chrysotile textile workers. I suspect, however, that the fibreexposure from this study may be overestimated, particularly among the portion of the cohort that iscomprised of miners and that a fair amount of misclassification took place in the amount of fibreexposure estimated for individual cohort members.

5.293 Gibbs and Lachance (1972) who published the initial exposure estimates for this population ofworkers stated that their cumulative dust concentrations for the cohort members may have been farfrom their actual experience. Subsequent dose response analysis of this cohort applied only a singleconversion factor to estimate fibre exposures from the dust count data for the entire cohort. In thestudy of Siberian chrysotile miners and millers (Tossavainen et al., 1999), those involved in miningexperienced average exposures of 0.08 f/cc, while those involved in two separate mills experiencedaverage exposures of 3.62 f/cc (ranges of average exposures for different millingoperations 0.37-6.21 f/cc) and 0.65 f/cc (range of average exposures 0.20-1.26 f/cc). The gravimetricsampling results indicate a 5-fold difference in average exposure for miners versus millers, while thesampling results for fibre exposures (the exposure of concern) indicate a 45-fold difference in averageexposure concentrations between miners and millers. Taking into consideration the variation withinmining and milling jobs, the difference between exposures to workers involved in these jobs would beeven greater. The large difference in the fibre exposures between miners and millers that wasobserved in the Tossavainen et al. (1999) study adds further support to my concern that applying asingle conversion factor from dust samples in estimating chrysotile fibre exposures for miners andmillers is the most likely explanation for the slope of the dose response for lung cancer in theMcDonald et al. study being so different from the slope for lung cancer based on the studies ofchrysotile textile workers. Stayner et al. (1996) provide a summary of data indicating that the lungcancer risk from exposure to chrysotile in either experimental animals or in humans is similar to therisk from exposure to amphibole asbestos in these species.

5.294 With regard to asbestosis, Stayner et al. (1997) published a dose response analysis based uponthe updated study of chrysotile textile workers by Dement et al. (1994). Their analysis indicates anexcess risk of 2 deaths from asbestosis per 1000 employees exposed to 0.1 f/cc for an occupationallifetime exposure of 45 years, or 0.2 per cent from a cumulative exposure of 4.5f/cc-years. Twoadditional studies estimate similar risks of death from asbestosis in relation to cumulative chrysotileexposure. The study by Berry et al. (1979) estimates a risk of 1 per cent for those categorized as"probably having asbestosis" among chrysotile textile workers who were exposed to an estimatedrange of 0.3 f/cc to 1.1 f/cc for 40 years, or a cumulative exposure of 12-44 f/cc-years. For thosecategorized as "certified asbestosis" a 1 per cent excess risk of death was associated with 63 f/cc-yearsof exposure. Huang (1990) estimates a risk of asbestosis of 1 per cent associated with 22 f/cc-years ofexposure during the manufacturing of chrysotile textile and friction products. Data from Dement et al.(1994) allow for an estimate of 2 per cent asbestosis associated with 22.5 f/cc-years of exposure.Studies of workers exposed to a mixture of chrysotile and crocidolite asbestos in the manufacture ofcement products have demonstrated an elevated the risk of "certified asbestosis" of 1 per centassociated with 10 f/cc-years of exposure (Finkelstein 1982). Finkelstein was of the opinion that heidentified a 1 per cent excess risk of asbestosis related to a lower cumulative dose of asbestosexposures than Berry et al. (1979) because of the longer period of follow-up of the cohort which gavemore time for the asbestosis to clinically manifest itself.

5.295 I am not aware of evidence that similar cumulative exposures to amphibole forms of asbestoswill result in a greater risk of asbestosis. Thus, it is difficult to make any distinction in potencybetween the amphiboles and chrysotile asbestos in relation to asbestosis.

WT/DS135/RPage 313

5.296 Chrysotile exposures related to numerous jobs and occupations has been associated withmesothelioma through epidemiological studies and case reports. In some situations, standby exposureonly was associated with mesothelioma. Based on epidemiological studies, the potency of chrysotileto induce mesothelioma may be less than that from other forms of asbestos. However, the rarity ofmesothelioma in the general population and the difficulty in determining asbestos exposure levelsexperienced by cohort members decades before measurements were taken, coupled with conversion ofdust counts in particles to fibres per cc, make it difficult to determine differences in potency estimateswith regard to the various forms of asbestos and mesothelioma. Based on toxicological study results,in terms of the quantity of dust deposited and retained in the lungs, chrysotile may be more potent thanother forms of asbestos in the induction mesothelioma and fibrosis (Wagner et al., 1974). Thepopulation attributable risk of contracting mesothelioma from chrysotile, however, will be greater thanfrom other forms of asbestos because of the much greater potential for exposure to chrysotile.

5.297 I have not seen any quantitative data related to decrements in lung function for eitherchrysotile and amphibole asbestos. On the basis of mortality from asbestosis, I assume there is littledifference in lung function related to the various forms of asbestos.

5.298 In summary to this question, in evaluating the epidemiological evidence, I see no basis forconcluding that the overall disease potential from exposure to the amphiboles is any different than thatfrom exposure to chrysotile asbestos with the possible exception that amphiboles may be more potentin causing mesothelioma and chrysotile may be more potent in causing lung cancer. Studies inexperimental animals demonstrate the ability of chrysotile asbestos as well as the amphiboles toinduce fibrosis, lung cancer and mesothelioma. From a public health standpoint, in terms ofquantification of disease, it would be extremely difficult to make any distinction between exposure tothe amphiboles and chrysotile asbestos fibres.

Dr. Musk:

5.299 Broadly it is my understanding that the relative pathogenicity of the different fibres for thevarious diseases is different (see Table)

CROCIDOLITE AMOSITE ANTHOPHYLLITE CHRYSOTILE

Asbestosis 1 1 1 1

Lung cancer 10 10 10 <1

Mesothelioma 100 10 5 1

Benign asbestospleural effusion/diffuse pleuralthickening

100 10 10 1

Pleural plaques 1 1 10 1

Question 4:

The parties in this dispute disagree as to the risk to human health associated with chrysotileasbestos fibres at low levels of exposure, i.e. either prolonged exposure to low concentrations offibres or occasional peaks of exposure. The European Communities considers that, because of alack of data at low levels of exposure, it is appropriate to endorse the linear relationship model toassess risks associated with such low levels of exposure. On the other hand, Canada is of the viewthat, at such low levels of exposure, empirical evidence suggests that there is a practical thresholdbelow which chrysotile asbestos fibres present no measurable effects on health.

WT/DS135/RPage 314

4.(a) Are epidemiological data available for low levels of exposure to chrysotile fibres and whatdo they show?

Dr. de Klerk:

5.300 There have been several epidemiological studies that have shown no increased risk for lowlevels of exposure to chrysotile, particularly in friction products industries.

Dr. Henderson:

5.301 So far as I am aware, there are no exposure-response data for such levels of exposure.

5.302 For example, EHC 203 states the following:

"Few data on concentrations of fibres associated with the installation and use of chrysotile-containingproducts are available, although this is easily the most likely place for workers to be exposed"[EHC 203, p 3.].

"Overall, the available toxicological data provide clear evidence that chrysotile fibres can causefibrogenic and carcinogenic hazards to humans. The data, however, are not adequate for providingquantitative estimates of the risk to humans. This is because there are inadequate exposure-responsedata from inhalation studies, and there are uncertainties concerning the sensitivities of the animalstudies for predicting human risk" [EHC 203, p.7].

"There is evidence that fibrous tremolite causes mesothelioma in humans. Since commercial chrysotilemay contain fibrous tremolite, it has been hypothesized that the latter may contribute to the induction ofmesotheliomas in some populations exposed primarily to chrysotile. The extent to which the observedexcesses of mesothelioma might be attributed to the fibrous tremolite content has not been resolved"[EHC 203, p 8-9].

"Epidemiological studies that contribute to our understanding of the health effects of chrysotileconducted to date and reviewed in this monograph have been on populations mainly in the mining ormanufacturing sectors and not in construction or other user industries. This should be borne in mindwhen considering potential risks associated with exposure to chrysotile" [EHC 203 , p 137].

Dr. Infante:

5.303 One means of determining the risk from low exposure levels for carcinogens is to estimate therisk from studies where exposure information is of reasonably good quality and the estimates of riskwere made using sound epidemiological principles and methodology. Once these studies have beenidentified, an appropriate way to determine quantitative risk from low exposure levels is to use all ofthe available data in a particular study and to estimate dose response. As mentioned in my response toQuestion 3(c) above, several studies of workers exposed to chrysotile asbestos can be used to estimaterisk from low levels of exposure. The studies mentioned above, which represent three separatepopulations of chrysotile textile workers demonstrate an excess relative risk of lung cancer rangingfrom 0.5-3 per cent for each fibre/cc-year of exposure. Risk assessment based on the study ofSouth Carolina chrysotile textile workers (Stayner et al., 1997) indicates that individuals exposedto 0.1 f/cc-year for an occupational lifetime of 45 years, e.g., 4.5 f/cc-years cumulative exposure tochrysotile, have an elevated risk of 5 extra deaths from lung cancer and 2 extra deaths from asbestosisper 1000 workers. Dose response for mesothelioma could not be estimated because there were toofew deaths from this cause in the study.

5.304 Epidemiological study results and case reports indicate that a large number of jobs whichentail occasional peak exposures to chrysotile asbestos have resulted in workers being diagnosed with

WT/DS135/RPage 315

mesothelioma. Furthermore, mesothelioma has been diagnosed among household contacts of asbestoscement workers (Magnani et al., 1992; Ascoli et al., 1996), among individuals living near chrysotilemining and milling operations (Began et al., 1992), or living in houses constructed with asbestoscement (Ascoli et al., 1996), or who were exposed to low cumulative amounts of chrysotile asbestosfrom brake lining work (Woitowitz and Rodelsperger, 1991), or from standby exposure as bakers(Ascoli et al., 1996). This information contributes to the evidence that very low exposure to all formsof asbestos can induce cancer. These observations of mesothelioma should be considered as sentinelevents for the pathologies other than mesothelioma that are more difficult to identify among largepopulations that experience relatively more remote exposure to chrysotile asbestos. In my opinion,these studies constitute high level risk from low levels of exposure to chrysotile asbestos.

Dr. Musk:

5.305 It my understanding there are epidemiological studies which do not show a statisticallyincreased risk of disease from low levels of exposure to chrysotile. However, the absence ofdemonstrating an increased risk does not mean that there is not some risk as it is not possible to provea negative and a threshold has not been demonstrated for any carcinogen (nor in my opinion is itbiologically likely that one exists).

4.(b) Is there a threshold below which exposure to chrysotile fibres does not induce (i) asbestosis,(ii) lung cancer, (iii) mesothelioma, and (iv) other asbestos-related pathologies, such as pleuralplaques? If there is such a threshold, is it a practical one or is it scientifically established?

Dr. de Klerk:

5.306 It is extraordinarily difficult to demonstrate lack of an effect, or a threshold effect, inepidemiological studies because of the ubiquitous problems of bias, confounding and chance. Inparticular, the smaller effect that needs to be demonstrated, the larger the study needs to be, both inpopulation size and follow-up time and such studies can rarely be done, even with animals.

Dr. Henderson:

(i) Asbestosis

5.307 Please see my answer to Question 3(b).

(ii) Lung Cancer

5.308 Please see discussion in Section C.1.(i). Please see also the statement from EHC 203 belowon the question of a threshold for carcinogenesis by asbestos. Some authorities favour a linear no-threshold model, whereas others argue that a threshold probably exists; nonetheless, there is nogeneral agreement on a numerical threshold for asbestos-induced lung cancer. Dement et al. [171]observed odds ratios of > 2.5 at 2.7-6.8 fibre-years of exposure among South Carolina asbestos textileworkers.

(iii) Mesothelioma

5.309 As indicated in previous discussion in this report, a linear dose-response model has beenidentified for mesothelioma induction by the amphiboles, and the dose-response relationship ismaintained at low occupational levels of exposure that overlap with environmental exposures:e.g. definite dose-response relationship reported by Rödelsperger et al. [25, 137] at asbestos fibreconcentrations in lung tissue of 100,000-200,000 fibres per gram dry lung tissue, with an indicationthat this relationship is maintained at lower levels of 50,000-100,000 fibres per gram dry lung (the

WT/DS135/RPage 316

level of 100,000-200,000 fibres corresponds to a cumulative dose of 1-2 fibre-years). Iwatsubo et al.[136] identified an increase in the relative risk at 0.5-0.99 fibre-year. No threshold has been identifiedfor amphibole-related mesothelioma. For chrysotile exposures, a dose-response relationship has alsobeen identified at high exposures, but to the best of my knowledge, there are no dose-response data forlow-level exposures to chrysotile.

5.310 On this point, EHC 203 states that "No threshold has been identified for carcinogenic risks"[for chrysotile; p 144]. At the same time, no increase in risk for mesothelioma has been identified atvery low-levels of exposure, of the type associated with well-maintained asbestos in place, in publicbuildings. However, it is impossible to ascertain whether or not there is an increase in risk at thisorder of exposure, because no control or reference group can be assembled where there is no asbestoscontent in lung tissue. If a threshold exists, it must lie somewhere in this area, between no exposure,low-level environmental exposure, and low-level occupational exposure.

5.311 De Klerk [115] has also commented on the difficulty or impossibility of distinguishingbetween background versus environmental mesotheliomas:

"There have been increases in the incidence rates of malignant mesothelioma in women, in thosewithout identified exposure to asbestos and, possibly, those younger than 35 years of age in Australiaand Western Australia. Although part of the first two increases, at least, may be attributable to specificexposure to asbestos, mathematical modelling of the Western Australia data suggests that there hasbeen about a twofold increase in incidence rates from the 1970s to the 1980s that may be due toincreased general environmental exposure to asbestos. ... The excess of 1 per million person yearsover this presumed 'background' rate is also, coincidentally, the amount that was estimated as possiblycaused by exposure of school children to 1 fiber per liter ... , a level that might result from use ofasbestos-based insulation or other general contamination of the environment with asbestos.

A final consideration in the use of national trend data for estimating environmental effects is thecomparison of the likely extrapolated risk from occupational data with the background risk estimatedhere or from Peto's Los Angeles data. From the Peto paper ... , the incidence of mesothelioma is relatedto age in the following way:

Incidence = 1.7 • 10-12 • (age)3.5

which translates to a lifetime risk to age 80 of just over 100 per million people, which is much greaterthan any of the estimated environmental risks described earlier. An equivalent risk from the adjustedWestern Australian data is about 160 per million lifetimes. The question remains as to how thesebackground risks and environmental risks interact. Is the postulated environmental incidence alreadyincluded in the background incidence, or should the risks be added or even multiplied together? Thisquestion is almost certainly unanswerable using epidemiological methods. ...

It is doubtful whether epidemiological methods ... could ever be definitive in deciding whether there isan appreciable hazard from general environmental exposure to asbestos ... or, more importantly,whether the hazard is large enough to justify specific remedial action ..." [pp 29-31].

5.312 Clearly, this issue is one major focus of dispute between experts; from the precedingdiscussions, it is my perception that — with the exception of asbestosis — no threshold has beendelineated, and that even those who claim that a practical threshold must exist cannot delineate such athreshold in precise numerical terms (in this respect, I do not know what the expression "practicalthreshold" really means).

Dr. Infante:

5.313 Thresholds have not been demonstrated for any substance known to cause cancer and there isno theoretical basis to assume a threshold for the diseases related to chrysotile, or other forms of

WT/DS135/RPage 317

asbestos, particularly when the mechanisms involved in the pathology are not fully understood.Furthermore, it is not possible to determine thresholds from epidemiological studies because of lack ofstatistical power to distinguish that the risk is virtually zero. [Note: At times, some investigators statethat a single point estimate from the lowest dose evaluated in an epidemiological study that does notdemonstrate a significant elevation in cancer risk constitutes a threshold level for the carcinogen.Such a conclusion is scientifically invalid. When estimating dose-response, one has more confidencein the risk related to a particular dose level by using all of the data available in the study. Using only asingle point estimate results in more instability in the estimate of risk for that data point in contrast tousing all of the available data in the study.]

5.314 Dose response analyses and modeling specifically for chrysotile asbestos exposure and lungcancer and asbestosis have been conducted recently by Stayner et al. (1997) using data from theDement et al. (1994) study. Alternative exposure-response models were evaluated as part of thestudy. A model designed to evaluate evidence of a threshold also was fitted for asbestos exposure inrelation to lung cancer and asbestosis. There was no significant evidence for a threshold in modelspertaining to either lung cancer or asbestosis.

5.315 With regard to mesothelioma, and other asbestos-related diseases, I am not aware of anyevidence of a threshold pertaining to either chrysotile, or other forms of asbestos. Furthermore, from apractical standpoint, even if there were a threshold for the chrysotile related diseases, the exposuresthat workers will routinely encounter in the future through continued use of chrysotile in commerce,will expose them to concentrations of asbestos that have already been related to pathology in humans.In other words, continued use of asbestos will continue to expose individuals to levels and exposurecircumstances that have already been related to disease. The threshold question, therefore, seemsmoot.

Dr. Musk:

5.316 It is my understanding that a threshold for disease has not been scientifically established.

4.(c) Is the linear relationship model an appropriate method for assessing the risk to humanhealth posed by exposure to chrysotile asbestos at low levels of exposure?

Dr. de Klerk:

5.317 The linear relationship model is generally used as a so-called "conservative" estimate, that is,if it is incorrect, it is more likely to err on the side of safety. In some ways, how one extrapolates riskassessment outside the range of available data is more of a societal decision than a scientific one.Biological plausibility could probably be given to any model.

Dr. Henderson:

5.318 In the absence of alternatives because of unavailability of data on exposure-responserelationships at low levels of exposure to chrysotile, the linear relationship model is widely employed.Under these circumstances, this may be an appropriate method for risk assessment at low levels ofexposure. Whether or not it is a valid method is unknown.

5.319 NICNAS 99 (p. 72) observes that:

"There are many problems associated with low-dose risk extrapolation, such as the assumption of alinear relationship. However, as insufficient data exist to indicate a threshold exposure for effect, thelinear extrapolation methodology provides a conservative worst-case scenario estimate of risk. Other

WT/DS135/RPage 318

confounding factors in estimating risks from epidemiological data are possible contamination by otherfibre types and inaccurate estimates of historical exposures."

5.320 Clearly, this too is one major focus of dispute between experts.

Dr. Infante:

5.321 A linear relationship model is appropriate for determining dose response for chrysotileexposure and lung cancer, and perhaps asbestosis and mesothelioma as well, but the most reasonablemodel for the latter two diseases are less clear than for lung cancer. With regard to this issue, Stayneret al. (1997) evaluated exposure-response relationships for chrysotile asbestos, and lung cancer andasbestosis by applying several alternate models. The exposure-response relation for asbestos and lungcancer gave the best fit when using a linear model. This observation is consistent with the conclusionsof other investigators, who have evaluated dose response for chrysotile asbestos textile workers, orother asbestos workers and mortality from lung cancer (McDonald et al., 1983; Peto et al., 1985;Enterline, Hartley & Henderson, 1987). Furthermore, there appears to be a linear relationshipbetween asbestos exposure and lung cancer over a wide range of exposures where such data areavailable. Therefore, it seems reasonable to accept a linear relationship for lung cancer whenextrapolating risks to exposures below the ranges that have been evaluated in epidemiological studies.Moreover, I am not aware of any literature that convincingly proves that the dose response forasbestos and lung cancer is non-linear. Thus, in my opinion, the linear model is the most appropriatemodel for estimating dose-response for chrysotile exposure and lung cancer.

5.322 A linear relationship might also be used for chrysotile asbestos exposure and asbestosisalthough one might make the argument that a non-linear model is also appropriate for asbestosis.Stayner et al. (1997) evaluated this issue using data from the Dement et al. (1994) study andconcluded that the association between chrysotile exposure and asbestosis appeared to be non-linear.Stayner et al. (1997) used a non-threshold, non-linear model and the estimates of asbestosis predictedfrom the model seem to fit very closely with point estimates for asbestosis from other studies ofchrysotile exposed populations as mentioned in my responses to Questions 3(c) and 4(a).

5.323 An analysis by Peto et al. (1985) of chrysotile asbestos textile workers shows that a linearmodel fits the data for mesothelioma with the cube of time since first exposure. In this non-thresholdmodel, the response is linear with dose of asbestos, but exponential with time since initial exposure.The predicted number of mesotheliomas by dose and time since first exposure was in reasonableagreement with the observed number. According to the authors, however, there were too few cases totest the model stringently and they did not attempt to fit other models to their data. Nevertheless,given the consistent observations of the long latency period between initial exposure to various formsof asbestos and the clinical manifestation of mesothelioma, it seems reasonable to use a model that islinear with exposure and exponential with time from initial exposure for chrysotile asbestos andmesothelioma.

Dr. Musk:

5.324 It is my opinion that the linear relationship model is the most appropriate one.

4.(d) Are there scientifically acceptable methods other than the linear relationship model whichcould be used to assess the risk to human health at low levels of exposure? What results do theysuggest?

WT/DS135/RPage 319

Dr. de Klerk:

5.325 While a threshold model suggests a lack of risk below a fixed level, it is unlikely that this riskwould be completely zero, so that if applied to a much larger population, such a risk could lead tocases of disease.

Dr. Henderson:

5.326 I am not aware of any other methods that have met with broad scientific acceptance or aconsensus. It has been suggested that an S-shaped curve might be more appropriate, but I have notseen any data on what the form of the S-curve might be; in other words, the S-shaped model appearsto presuppose the existence of a threshold, but no such threshold has been established to the best ofmy knowledge.

5.327 The problem with arguing that there exists a practical threshold level for lung cancer andmesothelioma induction is that it is impossible to delineate such a threshold in numerical terms,because of a lack of observational data. (Please see also my answer to Q.5(c)).

Dr. Infante:

5.328 From the public health perspective, it has been the convention to use non-threshold linearmodels for estimating cancer risk to humans. This is particularly the case for substances known tocause cancer in humans. One might deviate from this concept if the mechanism(s) by which thesubstance causes the cancer were known. This is not the case with chrysotile, or any other form ofasbestos. In the particular case of chrysotile asbestos, Stayner et al. (1997) selected several Poissonregression models to explore the shape of the exposure-response relationship between chrysotileasbestos exposure and risk of death from lung cancer. The models were capable of reflecting a widerange of exposure response patterns, including linear, sublinear and supralinear relationships. Theyalso considered a threshold model to determine whether there was evidence that exposures below acertain exposure concentration were equivalent to zero, i.e., that a threshold was present. Asmentioned above, for lung cancer, a linear model gave the best fit; for asbestosis, the responsepreferred was that based on a non-linear, non-threshold model. In both cases, the models did notprovide any support for the existence of a threshold. Thus, in my opinion, these models areappropriate to assess risk for these diseases as a result of occupational exposure to chrysotile asbestos.With regard to lung cancer and asbestos, I am not aware of any public health organization, orgovernmental agency that has ever used a non-linear model to estimate risk. During the hearings heldby the U.S. Occupational Safety and Health Administration (OSHA) as part of its rulemaking relatedto the standard promulgated for asbestos in 1994, numerous scientists were of the opinion that a non-threshold linear model was the preferred model to use for estimating the relationship between asbestosexposure and lung cancer. It is my opinion that a non-linear model is not an acceptable model to usein estimating dose response for asbestos exposure and risk of death from lung cancer. Formesothelioma, I would favor a non-threshold model that incorporates a linear relationship withexposure.

Dr. Musk:

5.329 I believe that there is no reason to discard the linear model as no threshold for any carcinogenis known to exist.

4.(e) To what concentration of chrysotile fibres and for how long must a person be exposed inorder to be considered at risk of developing a chrysotile asbestos-related disease (lung cancer,mesothelioma or other asbestos-related pathology)?

WT/DS135/RPage 320

Dr. de Klerk:

5.330 A person is "at risk" of developing a chrysotile asbestos-related disease after any exposure tochrysotile asbestos, the lower the amount of exposure, the lower the risk. For example, it could beestimated that there was a 50-50 chance that exposure to 1 fibre of crocidolite could cause 1 case ofmesothelioma among the whole population of the world (including all those who have ever lived),i.e. a very small probability, but still greater than zero.

Dr. Henderson:

5.331 This question iterates the issue of a threshold exposure. The answer is essentially the same asfor Questions 4(a)-4(d) in the absence of exposure-response data at low levels of exposure.

Dr. Infante:

5.332 The answer to this question depends upon the amount of risk that is considered unacceptableby a particular country. It is a matter of health policy. In the United States, the EnvironmentalProtection Agency (EPA) regulates risk to a level below one extra death in a population of 100,000people over its entire lifetime. I have already provided estimates for excess risk of death from lungcancer and asbestosis from chrysotile exposure. These risk estimates, however, are average risks to agroup of individuals that are based on maximum likelihood estimates (MLEs) and they do notincorporate statistical uncertainty in terms of variability, e.g., they are not based on upper 95 per centconfidence limits as is usually the health policy when estimating adverse health effects to a group ofindividuals at risk from exposure to an environmental insult. In addition, the risk estimates may onlybe appropriate to workers' health risks. They are derived from a group of healthy adults, who wereable to pass a physical in order to gain employment. They are not representative of individuals in thegeneral population who may be exposed to asbestos and have a compromised immune system, or beexposed to other conditions which may exacerbate their risk of contracting the various diseases relatedto chrysotile asbestos as estimated from a healthy worker population. There will always be some riskfrom exposure to asbestos and the degree of that risk will depend upon the amount of asbestosexposure in relation to the susceptibility of the individuals exposed in terms of their health status andother factors that interact to produce clinical manifestation of disease.

5.333 It is noteworthy, as mentioned in my response to Question 1(a) above, that the populationsurrounding the Quebec mining and milling operation (Begin et al., 1992) that was exposed tobackground levels of chrysotile asbestos, developed mesothelioma at an incidence of 62.5 cases permillion population per year, or 0.625 per 10,000 per year. This risk level translates to 0.5 per cent ofthe population developing mesothelioma over an 80-year lifetime from this background exposure.This estimate may well represent an underestimate of risk since the identification of cases was basedon a workman's compensation board review, and additionally, out-migration of inhabitants would alsoresult in loss of some cases. In the same report, it was pointed out that the largest increase in themesothelioma rate in Quebec was among individuals that had occupations where their exposure wouldbe occasional only, and that 33 per cent of these cases were exposed for less than a 5-year period.When one adds to this information to additional cases of mesothelioma reported in the literature thatare associated with standby exposure to chrysotile, it leads one to the conclusion that occasionalexposure for a short period of time, or constant low level exposure to chrysotile asbestos leads todeath from mesothelioma (and lung cancer and asbestosis). [Note: while an excess of lung cancer wasnot identified in the Camus et al. (1998) study of women in the same Quebec population, the studyhad limited power to detect an excess of lung cancer; it did, however, demonstrate an excess ofasbestosis and mesothelioma even though out-migration may have resulted in the loss of all three ofthese diseases in the study.]

WT/DS135/RPage 321

Dr. Musk:

5.334 In my opinion any level of exposure to chrysotile (or other form of asbestos) constitutes somerisk and that the level of "acceptable risk" is not a scientific issue but an issue for society to debate anddetermine at different times according to the evidence as they perceive it.

Question 5:

Canada states that, with controlled use, "health risks associated with occupationalexposure throughout the life-cycle of chrysotile asbestos can be reduced to acceptable levels alreadyrecognized as such by competent international organizations. The European Communitiesquestions this assertion and says that "les données scientifiques disponibles montrent quel'utilisation dite "sécuritaire" de l'amiante chrysotile ne permet pas d'empêcher un grand nombrede cas d'exposition entraînant des pathologies mortelles". ["available scientific evidence showsthat so-called "controlled" use of chrysotile asbestos does not make it possible to prevent manycases of exposure causing fatal pathologies"]

5.(a) Is there a generally agreed methodology applicable to any use of chrysotile-cement productsand other high-density chrysotile products throughout their life-cycle that can be referred to as"controlled use"? Is it embodied in international standards?

5.335 This is rather outside my areas of expertise. It does however appear theoretically feasible butpractically very unlikely given the problems with "downstream" use described above.

Dr. Henderson:

5.336 In principle, regulation and control of chrysotile and high-density chrysotile products isfeasible at some points of the life-cycle (manufacture and disposal), but in reality not others (pleasesee following discussion).

5.337 The manufacture of high-density products is usually carried out under closed conditions withdust extraction. As one example, the manufacture of chrysotile friction materials in Australia involvesthe following processes: following transfer of other ingredients required for the product mix,unopened 50 kg plastic bags of raw chrysotile are placed in the mixer and opened under dustextraction. The empty bag is then delivered into a second plastic bag attached to the mixer. Whenfull, this second bag is sealed and taken to a controlled disposal site. Mixing is a closed process.After mixing, the material is emptied under dust extraction before decanting into smaller buckets forweighing and use in moulding and finishing processes. The moulding is a hot process and whencomplete, the moulded product undergoes finishing processes that include grinding, grooving anddrilling — all carried out under dust extraction. The finished disc pads and commercial vehicle brakeblocks and linings are then wrapped and packed into sealed containers.

4.338 The potential for exposure includes opening and emptying chrysotile bags into the mixer, themoulding and finishing processes, and handling of damaged bags containing raw chrysotile. Theworkforce amounts to a few hundred workers, and the maximum exposures per employee vary fromminimal, to the largest group involved in the processing operations. Airborne asbestos fibre levels areassessed by personal monitoring. About 84 per cent of 461 samples between 1992 and 1997were < 0.1 f/ml; 10 per cent ≥ 0.01–≤ 0.2 f/ml; 6 per cent were ≥ 0.02–< 0.5 f/ml, and < 1 per centwere ≥ 0.5 f/ml. The manufacture of compressed asbestos fibre sheeting — most for export and theremainder processed into finished cut gaskets for industrial applications is also a closed processcarried out under similar conditions to the manufacture of friction products (please see NICNAS 99,pp 32-34). A total of 232 personal samples between 1991 and 1996 showed similar low airborne fibre

WT/DS135/RPage 322

concentration (58 per cent < 0.1 f/ml and only one sample ≥ 0.5 f/ml). Static samples recorded duringguillotine and trimming activities were all ≤ 0.05 f/ml.

5.339 Although controlled use of this type is feasible for these manufacturing processes, and fordisposal of materials left over (e.g. empty polythene bags), it is my perception that, historically and inreality, it is almost impossible to extend analogous controlled and regulated use to the end-users ofasbestos products such as workers involved in building construction and demolition (e.g. builders'labourers, carpenters, electricians, painters, plasterers and plumbers), or to individuals who carry outmaintenance or renovation work on their own homes, or to brake mechanics (please see AMR 99).This is because these groups add up to a large population of disparate and varied workers; many suchindividuals work for small businesses or are self-employed, so that it is difficult or impossible toextend controlled use or training to all of them.

5.340 Some of the supporting documentation submitted to the WTO refers to ILO recommendations,but it is also worth emphasizing that prohibition or regulation of asbestos-containing products variesfrom one nation to another, with different upper limits of airborne fibre concentrations (e.g. < 1 f/mlor < 0.1 f/ml). These are summarized in Tables 27 and 28 and Appendix 7 in NICNAS 99.

5.341 From the literature cited throughout this report and the reasons discussed, it is my perceptionthat broad agreement exists among experts that controlled use of chrysotile (or other varieties ofasbestos) is not a feasible option in the real world for certain worker groups, notably those involved inconstruction trades (e.g. see EHC 203).

Dr. Infante:

5.342 I am not familiar with any "agreed upon" methodology applicable to chrysotile-cementproducts and other high-density chrysotile products related to "controlled use" in the sense that theseproducts could be used without harm to human health. Perhaps the consensus is that when usingasbestos, the exposure should be controlled, and various countries have developed programmes orstandards that recommended, or require, specific engineering controls, work practices, training andeducation and personal protective equipment to control exposures to asbestos to the extent feasible.This, to me, is different from the concept of "controlled use" in the dialogue being used by Canada,which seems to imply that using or manipulating asbestos or asbestos containing products can be donein such a manner that people are not exposed, or that the risk from such exposure is de minimus.

5.343 I also am not aware of international standards related to "controlled use" of asbestos products.What you may be referring to here are recommendations from international organizations, orrecommendations or regulations (in the context of being enforceable by law) from various countries.These documents, however, should not be considered as international standards that lead to the"controlled use" of asbestos. For example, in 1994, the United States, promulgated a new standard foroccupational exposure to asbestos which required the permissible exposure limit (PEL) to be nohigher than 0.1 f/cc as an 8-hour time-weighted- average (TWA). In the opinion of OSHA, workersexposed to this PEL over an occupational lifetime (45 years) are still at a significant risk of developingasbestos related diseases. Therefore, the Agency included in the standard several provisions ancillaryto the PEL. One could argue that the PEL plus the ancillary provisions constitutes one of the bestexamples of the concept of "controlled use" of asbestos as I understand it in the concept being broughtforward by Canada. Yet, in the United States, the PEL for asbestos as well as the ancillary provisionsthat include training and education about the health hazards of exposure, work practices, requirementsfor personal protective equipment, medical surveillance, etc. are not complied with for various reasonsin a large number of workplaces. Based on the observations of violations of several provisions of theasbestos standard in the United States, discussions with occupational safety and health personnel fromother countries and my review of the literature, it is my opinion that a controlled use concept forchrysotile asbestos is not realistic in workplace situations. It would be much less realistic as applied

WT/DS135/RPage 323

to non-occupational situations where individuals would make repairs involving the manipulation ofasbestos products in their homes.

5.344 I am not aware of any international standard that embodies "controlled use" of asbestos in thecontext that the manipulation of asbestos, or asbestos containing products will result in exposures thatwill not result in harm to many of those exposed. Furthermore, it should be recognized thatprogrammes to control asbestos in many countries are "agreements" and as such, they are notenforceable by law. Even more disconcerting is the observation that countries like the United Statesthat promulgated stringent requirements to control asbestos find that their standards are often violated.A general downfall with the concept of "controlled use" is that it relies upon human behaviour, whichcannot be controlled in too many situations. Hence, it is unreliable.

Dr. Musk:

5.345 In my opinion "controlled use" of asbestos is theoretically possible but not practicallyfeasible. The second half of the question is not within my area of expertise.

5.(b) To what extent is controlled use feasible in terms of the training of those involved in suchuse, implementing process changes, monitoring, etc.? Have there been studies conducted in thisregard and what do they show?

Dr. de Klerk:

5.346 This is definitely outside my area of expertise.

Dr. Henderson:

5.347 In theory, training of specific workers (e.g. products manufacture) and some other personnelin the controlled use of chrysotile ought to be feasible (as for other potentially hazardous materialssuch as radioactive materials used in nuclear reactors). As a matter of common sense, training is mostlikely to be effective when there is a small and cohesive workforce using materials that are notaccessible to most other workers who have limited or no training, and when the workers have a clearunderstanding of the hazards and risks of the materials handled.

5.348 However, training of this type becomes less feasible or impossible in practice when there is alarge and non-cohesive workforce and when there is general accessibility to the materials in question(e.g. builders' labourers, carpenters, electricians, plumbers and so forth at building sites, and brakemechanics).

5.349 Even so, I am sceptical about the consistent and universal effectiveness of trainingprogrammes in the real world, even when the workforce is small and cohesive, and involved in thehandling of materials that present a clear hazard (e.g. radioactive isotopes). For example, followingthe recent mishap at a nuclear reactor in Japan, a BBC report broadcast by ABC News Radio onSunday 31 October 1999 pointed out that the workers at the Japanese nuclear plant had been poorlytrained, with a poor understanding of the risks.

5.350 An analogous report was printed in The Guardian Weekly (October 28-03 November 1999,p. 9):

"Britain's key nuclear warheads factory this week admitted to more than 100 breaches of safety in thepast year but ... the Director of Communications at the Atomic Weapons Establishment inAldermarston ... branded claims that only luck had prevented an accident worse than that in Japan as'irresponsible scaremongering'. ... The Plant's emphatic denials came after The Observer newspaper

WT/DS135/RPage 324

published details of a leaked report highlighting more than 100 dangerous incidents since Septemberlast year. ... Among these were eight breaches of the 'criticality' rules ... and eight instances ofenvironmental contamination outside the site. There were also eight occasions when materials —including plutonium — were incorrectly packaged or labelled, and 19 highly serious health and safetyincidents, including all fire-fighting pumping appliances being unfit to service. ... The lead-in of thecatalogue of breaches comes as the environment agency prepares to decide whether to prosecute theprivatised Aldermarston for dumping tritium, a radioactive substance, in a stream from which Reading'sdrinking water is sourced ... it also follows the imposition of a £22,000 fine on the company after twoworkers breathed in radioactive particles from plutonium that had escaped from a laboratory in Augustlast year. ... The establishment insisted it was 'regulated up to the eyeballs' by external agencies andsaid it was penalised for its 'openness and transparency' in reporting breaches ...".

Dr. Infante:

5.351 Training is beneficial in reducing worker exposure to toxic substances in some circumstances,but minimizing exposure to asbestos to the extent that a significant amount of disease will not occur,is not one of these circumstances because of the extent of the training provisions necessary to reduceexposures, the lack of ability to reach all of the potentially exposed populations, and the wide-spreaduse of chrysotile asbestos. Training would be relatively better achieved in the manufacturing sectorwhere the employed population is relatively more stable as compared to the construction sectorwherein many workers are transient employees. Because of the transient nature of the workforce andthe cost involved to train workers, there is a tendency to not train those who will only be employed forshort periods of time.

5.352 In terms of studies related to the feasibility of using chrysotile asbestos in a "controlledmanner" OSHA health compliance data may offer some insight. The United States has an asbestosstandard that includes training requirements that are enforceable by law, and violations of which arepunishable by monetary penalties. Yet, improper work practices (presumably a reflection of lack oftraining) and violations of the permissible exposure limit continue to be identified. Since 1980,OSHA compliance officers have identified almost 14,000 violations among establishments for failureto comply with provisions of its asbestos standard. During the recent 3-year period of 1996-98,over 4,000 violations have been cited. Because of the small number of OSHA compliance staff inrelation to the number of facilities in the United States, it has been estimated that compliance officersare able to visit industrial workplaces an estimated one time in every 84 years. Thus, the non-compliance with provisions of the United States asbestos standard as identified by its complianceofficers, represents a "tip of the iceberg" in identifying non-compliance with provisions of thestandard that are thought necessary to control asbestos exposure in the workplace. As difficult as thissituation is in the occupational setting, I am inclined to believe that training leading to "controlled use"of chrysotile would be even more difficult to achieve outside of the occupational setting.

Dr. Musk:

5.353 As above in my opinion controlled use of chrysotile is theoretically feasible but probably notpractically possible. The second question part of the question is not in my area of expertise.

5.(c) Can controlled use, when properly applied, reduce exposure levels to chrysotile fibres tobelow 0,1 f/ml26? Can controlled use provide the assurance that there will be no peaks above thisfigure for any type of use of high-density chrysotile products? For workers or other personsexposed to chrysotile asbestos at this level, can you quantify the risk?

26Maximum exposure limit which was authorized in France before the ban.

WT/DS135/RPage 325

Dr. de Klerk:

5.354 See Table in paragraph 5.197 above.

Dr. Henderson:

5.355 When properly applied, controlled use in specific situations (e.g. friction productsmanufacture) can reduce airborne chrysotile concentrations to < 0.1 f/ml for most of the time: e.g. asmentioned in my answer to Question 5(a), at the Australian friction products factory in Victoria,84 per cent of 461 samples (1992-1997) showed an airborne fibre concentration < 0.1 f/ml; however,the remainder showed airborne fibre levels above this figure, although the concentrations were stilllow. This also demonstrates that one cannot always guarantee that fibre concentrations will neverexceed 0.1 f/ml, even in highly regulated circumstances such as the manufacture of high-densitychrysotile products. For the reasons discussed in preceding sections of this report, it is not possible toquantify the risks (e.g. lung cancer or mesothelioma) from occasional peak exposures > 0.1 f/ml,because these risks necessarily depend upon fibre type, the intensity of the exposure, and thefrequency and duration of exposure; in addition, quantification of the risk necessarily involvesextrapolation from a linear dose-response model, which is the subject of dispute, because there are noobservational data on risks from low-level chrysotile exposure.

5.356 Tables 12 and 13 above (see my response to Question 1(d)) give some estimates of lungcancer and mesothelioma risk at various levels of exposure to chrysotile.

5.357 Finally, occasional bursts or peak concentrations of fibres can be released from buildings bycatastrophic and uncontrollable events that include, for example, destruction by fire [244-247],earthquake, or explosions [246], including war (e.g. bombing of cities). Although disasters like theseare infrequent — and the fall-out from fires is probably of little consequence to nearby residents or thepublic in general [247] — they pose a potential health risk for some occupational groups likely toencounter asbestos fall-out and debris more often, such as fire fighters and those involved in clean-upoperations.

Dr. Infante:

5.358 Since I believe that "controlled use" is a misnomer in relation to potential for exposure toasbestos, I prefer to answer this question in the context of whether a stringent standard for the controlof asbestos in the workplace can reduce exposures to below 0.1 f/cc. In many situations, whenstandards are properly applied, or adhered to, it is possible to maintain exposures below 0.1 f/cc.However, in many situations today, even in the presence of a stringent standard to control exposure,the 0.1 f/cc level for asbestos is exceeded, specified work practices and housekeeping provisions thatare required to reduce exposures are not adhered to, communication of hazards is not provided, andappropriate personal protective equipment is not used. As a result, both average exposures and peakexposures above 0.1 f/cc will occur. With regard to high density chrysotile products specifically,over 2,000 violations of the OSHA asbestos standard have been identified in standard industrialclassification (SIC) sectors where potential for exposure to chrysotile asbestos would occur, namely:wrecking and demolition work, asbestos products manufacturing, roofing, siding and sheet metalwork, manufacturing of gasket, packing and sealing devices and automotive repair shops.

5.359 In addition to the above examples, in my opinion, many work practices are simply not goodenough in some situations to reduce exposures below 0.1 f/cc. For example, even though asbestoscement products may be "pre-sized" in manufacturing, cement sheets need to be modified, drilled, cutand ground to fit them into specific areas. Although wetting may be used in some of these situationsfor cutting asbestos cement, the material dries and the dust becomes airborne resulting in uncontrolledasbestos dust being released into the work environment. Even in such situations where workers are

WT/DS135/RPage 326

provided with respirators to prevent exposure, the respirators supplied for the situation are often thewrong ones. In addition, respirator effectiveness depends upon how well the masks fit the face, howoften they are replaced, cleaned and repaired and how well the employees are trained to select, cleanand repair their respirators. In spite of rules requiring good respirator practices, employers oftensimply do not comply with the regulations.

5.360 In a more general sense, non-compliance with regulations that could assist in the control ofasbestos to exposure levels below 0.1 f/cc is often the result of human error, (not recognizing thatasbestos is present, not understanding the requirements to protect workers from exposure in specificsituations), wilful non-compliance, poor judgement, accidents, inability to adequately reach thepotentially exposed population with proper training and education of the health hazard, etc. Based onmy experience in occupational health and the results of OSHA compliance data and the literature, it ismy opinion that it is impossible to use asbestos and assure that exposures can be kept below 0.1 f/cc ina great number of situations.

5.361 Regarding health risks from chrysotile exposure to 0.1 f/cc, as mentioned in my answer toQuestion 3, the risk assessment conducted by Stayner et al. (1997) based on the study by Dement et al.(1994) provides a good estimates of disease risks from exposure to chrysotile asbestos which are notmuch different from those provided from other studies. The analyses estimate that exposureto 0.1 f/cc of chrysotile asbestos for an occupational lifetime of 45 years, e.g., 4.5 f/cc-years ofcumulative exposure, results in five extra deaths from lung cancer and two extra deaths fromasbestosis per 1000 workers. The analysis did not include mesothelioma because there were too fewdeaths from this cause in the study to provide estimates of risk. I have not seen risk assessments formesothelioma based on chrysotile exposure. Thus, I cannot provide a quantitative estimate of risk forthis cause of death.

5.362 Of these diseases, the risk of developing asbestosis is most likely underestimated. Thisunderestimation is a reflection of the fact that most risk assessments are based on mortality studies andindividuals diagnosed with asbestosis often die from other causes. For example, in the study byFinkelstein (1982), of 24 individuals with certified asbestosis, who had died, 14 (58 per cent) diedfrom lung cancer or mesothelioma and three (12.5 per cent) died from ischemic heart disease. Thus,70 per cent of the individuals with asbestosis who died would not have been identified from amortality study of this population of exposed workers. Asbestosis may also be underestimated in astudy because it may be confused clinically with other non-malignant lung diseases. Therefore, therisk of death from asbestosis based on mortality studies may usually be underestimated. I am notfamiliar with quantitative estimates of risk from other pathologies related to chrysotile asbestos.

Dr. Musk:

5.363 The question is not within my area of expertise except that risks from exposure could becalculated if exposure levels are known.

5.(d) Is it possible to control the risks to human health presented by exposure to high-densitychrysotile products, in particular chrysotile-cement, throughout the life-cycle of the product? Iscontrolled use a feasible and practicable option in everyday life for workers who are exposedoccasionally to potentially high levels ("peaks of exposure") of chrysotile asbestos (such asplumbers, electricians, maintenance, repair, insulation, demolition, waste management and"handyman" type persons)?

Dr. de Klerk:

5.364 See my response to Question 5(a).

WT/DS135/RPage 327

Dr. Henderson:

5.365 From my perspective, these questions are crucial to the dispute before the WTO. In myopinion, the answer to both questions is NO. I do not see how asbestos in place — or existingchrysotile products — can be controlled at every point of end-use. For example, EHC 203 indicatesrepeatedly that exposure is most likely to affect workers involved in building construction ordemolition; this is because of the large number of these workers involved in myriad different tasks;in addition, these various workers represent a non-cohesive workforce, many self-employed orworking as part of "small business".

5.366 My perception is also based on studies such as that reported by Kumagai et al. [4] on Japaneseworkers involved in the repair of asbestos-cement pipes, where fibre concentration for fibres > 5 µmin length ranged from 92 f/ml inside the hole where this work took place (range 48-170 f/ml) and up to15 f/ml outside the hole; the final sentence of the Abstract for this Report indicates that only about 18per cent of the workers used a protective respiratory device. In addition, a survey in Finland foundoccasional high fibre concentrations inside personal protectors during asbestos removal work,suggesting that these devices are not always effective.

5.367 Another factor that merits consideration in some societies such as Australia, is poor workercompliance with controls. For example, I am aware of non-compliance in the use of protectiveequipment (despite penalties), because respiratory protective devices may be cumbersome anduncomfortable — especially in hot climates like Australia — with skin irritation from sweataccumulating within them.

5.368 From the literature cited throughout this report and the reasons discussed, it seems clear thatthere is a broad consensus among experts that controlled use of chrysotile (or other varieties ofasbestos) is not feasible in practice for certain worker groups, notably those involved in constructiontrades (e.g. see EHC 203).

Dr. Infante:

5.369 As mentioned in my responses to several previous questions, in my opinion, it is not possibleto control the risk to human health posed by exposure to chrysotile asbestos throughout its life cycle."Controlled use" is not a feasible and practical option in everyday life for workers. While "controlleduse" is relatively more achievable in the manufacturing sector, violations of regulations enforceableby monetary fines still occur. In the construction sector, control of exposure to asbestos is muchmore difficult to achieve as compared to manufacturing. Workers who have jobs as plumbers,electricians, maintenance personnel, repairmen, insulators, demolition, waste management andhandymen will most likely experience intermittent peak exposures to asbestos. These exposures resultfrom lack of awareness of the hazard, lack of recognition of the hazard, lack of personal protectiveequipment, lack of training on the maintenance of the protective equipment, etc. as mentioned inresponses to other questions above.

Dr. Musk:

5.370 In my opinion controlled use is probably not practically possible but this is not an area of myexpertise.

5.(e) Is it possible to control the risks to human health presented by exposure to high-densitychrysotile products, in particular chrysotile-cement, in non-occupational circumstances, such asintervention on these products by private individuals (cutting, sawing, removal, etc.)? Is controlleduse a feasible and practicable option for this category of the population?

WT/DS135/RPage 328

Dr. de Klerk:

5.371 See my response to Question 5(a).

Dr. Henderson:

5.372 As a follow-on to my answer to the preceding question, my answer to both of these questionsis also NO. However, the risks from occasional or infrequent interventions on chrysotile-onlyproducts (e.g. by home "handymen") - although not quantifiable because of absence of data - must bevery small for lung cancer and mesothelioma, and non-existent for asbestosis.

Dr. Infante:

5.373 As difficult as it is to control exposure to chrysotile asbestos during intervention with cementproducts in the occupational setting, it is much more so in non-occupational circumstances becausethere is no effective means of identifying the potential population at risk. As a result, there is a lack ofawareness of the hazard, lack of recognition of the hazard, lack of personal protective equipment, lackof training on the maintenance of the protective equipment, etc. as mentioned in responses to otherquestions above. Therefore, in my opinion, it is not possible to limit exposure in such circumstances.

Dr. Musk:

5.374 Controlled use is probably not practically possible, in my inexpert opinion.

Question 6:

The parties disagree as to the relative pathogenicity of chrysotile fibres vs. substitute fibres,in particular cellulose fibres, para-aramid fibres, glass fibres and polyvinyl alcohol (PVA) fibres.Canada considers that, overall, substitute fibres have not been demonstrated to be less toxic thanchrysotile fibres, and that, by banning chrysotile, France has replaced the "much studied butnonetheless undetectable risk associated with modern uses of chrysotile with the unknown, andperhaps greater risk associated with the use of substitute fibres" [Premier exposé oral du Canada,paragraph 90]. On the other hand, the European Communities argues that none of the substituteproducts -fibrous or non-fibrous- for chrysotile, and in particular none of the substitutes forchrysotile-cement, has been classified as a proven carcinogen to humans; hence, overall, substituteproducts present less of a risk to human health than chrysotile asbestos [see Deuxième soumissionécrite, pp. 10-15].

6.(a) Is it correct to argue that non-fibrous substitutes are safe or less hazardous than chrysotileand that concern over potential health risks should be focused on fibrous ones? In this context,could you elaborate upon the "effet fibre" ["fibre effect"] of substitute fibres? What generalconclusions can be drawn as to the respirability and biopersistence of substitute fibres?

Dr. de Klerk:

5.375 As outlined above, the pathogenicity of fibres is related to their size, shape, durability andquantity. Thus, all the parts to this question can be answered in the same way. The argument here iswhether it is safer to stick with the well-studied chrysotile that has a semi-quantifiable and definitecarcinogenic risk, than to use other substances which have the potential to increase risk in anunquantifiable way, ie. the "better the devil you know" principle. For example, para-amid fibres haverecently been classified by IARC in Group 3, that is, 'not classifiable as to its carcinogenicity'.

WT/DS135/RPage 329

5.376 Substitutes need to be compared to chrysotile in terms of the parameters listed above, namely,size, shape, durability and quantity. These are all properties of fibres and therefore "concern shouldbe focused on fibrous substitutes". Substitute fibres can then be compared with chrysotile on thefour parameters. I am inexpert in commenting on the "extent to which hazardous concentrations canbe controlled" but it is my understanding that all four substitutes mentioned involve less dustyoperations than equivalent ones involving chrysotile. As far as the other three parameters areconcerned: all four substitutes except glass fibre produce a larger proportion of non-respirable fibresthan chrysotile does, but respirable fibres are similar for all substances and glass fibre is the leastdurable; all four except cellulose are less durable than chrysotile, but cellulose is much less dusty andhas also been in use for a long while without evidence of ill effect.

5.377 On balance, the substitute fibres appear less likely to cause adverse effects (from their fibres)than chrysotile.

Dr. Henderson:

5.378 Current thinking on this issue indicates that the bio-hazards — specifically the carcinogenicrisks - of all fibres are determined by the three Ds: dose, fibre dimensions and durability (bio-persistence) [248-250]. Therefore, substitute materials that have engendered most concern are fibrousmaterials as opposed to non-fibrous substances (non-fibrous materials may or may not show differenteffects in terms of toxicology, but this discussion focusses on carcinogenic risks). For example,refractory ceramic fibres (RCF) are a cause for concern [251] because they may have dimensionssimilar to those of the amphibole varieties of asbestos and RCF have been reported to inducemesothelioma in experimental animals.

5.379 In a 1995 review, de Vuyst et al. [248] concluded that:

"The group of man-made mineral or vitreous fibres (MMMFs or MMVFs) includes glass wool, rockwool, slag wool, glass filaments and microfibres, and refractory ceramic fibres (RCFs). Experimentalobservations have provided evidence that some types of MMVF are bioactive under certain conditions.The critical role of size parameters has been demonstrated in cellular and animal experiments, whenintact fibres are in direct contact with the target cells. It is, however, difficult to extrapolate the resultsfrom these studies to humans since they bypass inhalation, deposition, clearance and translocationmechanisms. Inhalation studies are more realistic, but show differences between animal speciesregarding their sensibility to tumour induction by fibres. Fibre biopersistence is an important factor, assuggested by recent inhalation studies, which demonstrate positive results with RCF for fibrosis, lungtumours and mesothelioma. There is no firm evidence that exposure to glass-, rock- and slag wool isassociated with lung fibrosis, pleural lesions, or nonspecific respiratory disease in humans. Exposure toRCF could enhance the effects of smoking in causing airways obstruction. An elevated standardmortality ratio for lung cancer has been demonstrated in cohorts of workers exposed to MMVF,especially in the early technological phase of mineral (rock slag) wool production. During that period,several carcinogenic agents (arsenic, asbestos, polycyclic aromatic hydrocarbons (PAH)) were alsopresent at the workplace and quantitative data about smoking and fibre levels are lacking. It is notpossible from these data to determine whether the risk of lung cancer is due to the MMVFs themselves.No increased risk of mesothelioma has been demonstrated in the cohorts of workers exposed to glass-,slag- or rock wool. There are in fact insufficient epidemiological data available concerning neoplasticdiseases in RCF production workers because of the small size of the workforce and the relatively recentindustrial production" [abstract].

5.380 In a 1999 review published in French, Boillat et al. [250] came to similar conclusions:

"The group of man-made mineral fibres includes slagwool, glasswool, rockwool, glass filaments andmicrofibres, as well as refractory ceramic fibres. The toxicity of mineral fibres is determined by severalfactors such as the diameter (< or = 3-3.5 microns) and the length of the fibres (< 100 microns), theirbiopersistence, which is much shorter for man-made mineral fibres than for asbestos fibres, their

WT/DS135/RPage 330

physicochemical structure and surface properties, and the exposure level. The chemical composition ofthe various types of man-made mineral fibres depends directly on the raw material used to manufacturethem. While naturally occurring fibres are crystalline in structure, most man-made mineral fibres areamorphous silicates combined with various metal oxides and additives. Observations usingintracavitary administration have provided evidence that some types of man-made mineral fibres arebioactive in cellular and animal experiments and may induce lung tumours and mesothelioma. It isdifficult to extrapolate these results to humans since they bypass inhalation, deposition, clearance andtranslocation mechanisms. Inhalation studies show more realistic results but differences are observedbetween animal species regarding their sensibility to tumours. There is no firm evidence that exposureto various wools is associated with lung fibrosis, pleural lesions or nonspecific respiratory disease inhumans. A possible exception may be mentioned for refractory ceramic fibres. A slightly elevatedstandard mortality ratio for lung cancer has been documented in large cohorts of workers (USA, Europeand Canada) exposed to man-made mineral fibres, especially in the early technological phase. It is notpossible to determine from these data whether the risk of lung cancer is due to the man-made mineralfibres themselves, in particular due to the lack of data on smoking habits. No increased risk ofmesothelioma has been demonstrated in these cohorts. Epidemiological data are insufficient at this timeconcerning neoplastic diseases in refractory ceramic fibres" [abstract].

5.381 In one study on RCF, Glass et al. [252] reported that:

"In recent inhalation experiments conducted with both rats and hamsters ... at the highest dose tested ...there was an increased incidence of tumours in both species. Lower doses were only examined in therat and at these doses there was no significant excess of lung tumours. Epidemiological investigationsof workers engaged in the manufacture of ceramic fibres have shown a small excess of pleural plaques.This phenomenon is being further investigated but could be due to confounding exposures. Thepopulations available for study are small and their exposures fairly short, but it is considered prudentthat they should remain under surveillance for some time to come. This is despite the fact that presentexposures in the ceramic fibre industry are low (< 1 f/ml) and are being reduced" [abstract].

5.382 Okayasu et al. [253] also found that RCF-1 fibres were less cytotoxic and mutagenic thanchrysotile:

"Cytotoxicity and mutagenicity of tremolite, erionite and the man-made ceramic (RCF-1) fibre werestudied using the human-hamster hybrid A(L) cells. Results from these fibres were compared withthose of UICC Rhodesian chrysotile fibres. The A(L) cell mutation assay, based on the S1 gene markerlocated on human chromosome 11, the only human chromosome contained in the hybrid cell, has beenshown to be more sensitive than conventional assays in detecting deletion mutations. Tremolite,erionite and RCF-1 fibres were significantly less cytotoxic to A(L) cells than chrysotile. Mutagenesisstudies at the HPRT locus revealed no significant mutant yield with any of these fibres. In contrast,both erionite and tremolite induced dose-dependent S1- mutations in fibre-exposed cells, with theformer inducing a significantly higher mutant yield than the latter fibre type. On the other hand, RCF-1fibres were largely non-mutagenic. At equitoxic doses (cell survival at approximately 0.7), erionite wasfound to be the most potent mutagen among the three fibres tested and at a level comparable to that ofchrysotile fibres. These results indicate that RCF-1 fibres are non-genotoxic under the conditions usedin the studies and suggest that the high mesothelioma incidence previously observed in hamster mayeither be a result of selective sensitivity of hamster pleura to fibre-induced chronic irritation or as aresult of prolonged fibre treatment. Furthermore, the relatively high mutagenic potential for erionite isconsistent with its documented carcinogenicity" [abstract].

5.383 An important consideration is that fibre dimensions for some substitute materials(e.g. fibreglass) can be varied according to the manufacturing processes employed, so that they can bedesigned to have fibre characteristics and dimensions different from asbestos, or similar to asbestos:as one example, the dimensions of fibreglass can be varied and when implanted into experimentalanimals, fibres of the "right" size can induce mesothelioma.

5.384 For this reason, testing of substitute materials with fibre dimensions similar to those ofasbestos should be carried out before these materials are used in products available to the general

WT/DS135/RPage 331

public (e.g. testing for toxicology, clastogenicity, DNA strand breaks, mutagenicity and free radicalgeneration using in vitro systems and/or testing in vivo — such as the intraperitoneal test in rats) [248,249, 251-256].

5.385 Nonetheless, it is my perception that lumping all substitute fibres together is as erroneous aslumping amphibole and chrysotile fibres into the same category. For example, RCF are the subject ofcontinuing concern, but other substitute fibres such as cellulose fibres, para-aramid fibres andpolyvinyl alcohol (PVA) fibres appear to be different from chrysotile, in terms of fibre dimensionsand especially bio-persistence.

5.386 NICNAS 99 summarizes these considerations in the following terms:

"Any substitution of chrysotile should be with a less hazardous substance. There has been ongoingdebate regarding the health effects of alternatives, such as synthetic mineral fibres (SMF), naturalorganic fibres and synthetic organic fibres.

In general, less data on health effects of alternative materials (in comparison to asbestiform fibres) areavailable and because of this, it is difficult to make an assessment of the pathogenicity and potentialcarcinogenicity of many substitutes.

Although not the only determinant of potential pathogenicity, fibre dimensions (length, width andaspect ratio) are considered to be [some] of the most important factors associated with carcinogenic(lung cancer and mesothelioma) potential ... The commonly accepted 'peak hazard' dimensions ...are > 5 µm long (length) and < 3 µm wide (diameter).

The most commonly used alternatives in Australia (and overseas) for friction materials are aramidfibres, attapulgite, fibreglass, refractory ceramic fibres (RCF), semi-metallics, mineral wool, steel wool,cellulose, titanate fibres and wollastonite, and for gaskets are glass fibre, carbon fibre and aramid fibre.

... It should also be noted that ... differences in fibre length, diameter and surface properties may leadto entirely different toxicological profiles.

A recent report by EC concludes that the available data are generally supportive of the conclusion thatPVA, cellulose, p-aramid, glass wool and slag wool are likely to be safer in use than chrysotile.However, RCFs are the subject of ongoing concern ..." [p 125].

Dr. Infante:

5.387 I have not seen any information that indicates that non-fibrous substitutes for chrysotile arecarcinogenic, or cause non-malignant lung diseases. I would focus attention on the fibrous substitutesin terms of their ability to reach lung tissue (respirability) and their known toxicity. Clearly, if thesubstitute fibres are not respirable, there is little concern for their "potential" to cause lung diseases.(Attention would then focus on adverse effects from exposure to the skin and eyes.) If the substitutefibres are respirable, then attention needs to focus on their toxicity relative to that of chrysotile in theirability to cause lung cancer, non-malignant lung diseases and mesothelioma.

5.388 The data I have reviewed in this area of investigation appear to indicate that polyvinyl alcoholfibres (PVA) are mostly in the range of 10-16 microns in diameter and hence are too large torespirable and thus cause lung disease. In terms of biopersistence, if they were respirable, they woulddegrade very slowly. Para-aramid fibres are also generally 10-12 microns in diameter and they alsowould have little chance of being respired. These fibres, however, contain fibrils of about 0.2 micronsin diameter that can be liberated with high energy input and they would be respirable. P-aramid fibrilsgreater than 5 microns in length are less biopersistent than chrysotile fibres greater than 5 microns inlength (Searl, 1997). Data for dimensions of cellulose fibres show a median length and diameter ofabout 7.5 and 1.50 microns, respectively, which indicates that they are in the respirable range (Muhle

WT/DS135/RPage 332

et al. 1997). In terms of biopersistence, cellulose fibres had a mass half time in the rat lung of 72 daysand bioaccumulated in the lungs. Data on the distribution of glass fibres indicates that the majorityare in the respirable range, but the fibre size distribution of glass filaments indicates that a smallportion are in the respirable range. Glass fibres are less biopersistent than chrysotile fibres. Ingeneral, in terms of the combination of respirability and biopersistence, with the exception of cellulosefibres, it appears that the substitute fibres would have less bioaccumulation in the lung than chrysotilefibres because they are either less respirable, or they are not as biopersistent.

5.389 The role of biopersistence in relation to toxicity is complicated. Chrysotile fibres are lessbiopersistent than amphibole fibres, yet experimental data demonstrate a similar potency for lungcancer, mesothelioma and fibrosis.

Dr. Musk:

5.390 I agree with the Canadian argument philosophically. However there is no evidence that Iknow of carcinogenicity of substitutes in animal studies and only rockwool has been associated withincreased lung cancer risk in epidemiological studies.

6.(b) To what extent do physical characteristics and chemical properties of substitute fibresdetermine their toxicity? Is it correct to say that man-made fibre substitutes are superior to naturalfibre ones in terms of the extent to which exposure to hazardous concentrations can be controlledduring the various stages of production? Is your opinion based on one or more of the followingevidence: (i) chemical/physical characteristics of the substitute fibres, (ii) epidemiological data, (ii)in vitro evidence, (iv) in vivo evidence?

Dr. de Klerk:

5.391 See my response to Question 6(a).

Dr. Henderson:

5.392 These questions are covered to a large extent in my answer to the preceding question. Again,dose, fibre dimensions (including surface chemistry) and bio-persistence appear to represent theproperties that determine the toxicity and carcinogenicity of fibres of any type. The issue ofcontrollability during various stages of production is an engineering and industrial question, and fallsoutside my expertise.

5.393 My opinions concerning the potential bio-hazards of these fibres are based on the physicalcharacteristics of the substitute fibres, in vivo evidence (tumour induction in experimental animals)and in vitro studies (mutagenicity analogous that reported for other known carcinogens). To the bestof my knowledge, there are no large-scale epidemiological studies on cellulose fibres, para-aramidfibres or PVA fibres; two large epidemiological investigations on slag wool fibres in both Europe andthe United States did show an increase in the relative risk for lung cancer among the productionworkers, but this effect may have been explicable by other confounding factors involved in themanufacture of these materials.

Dr. Infante:

5.394 As a matter of general toxicology, I would focus concern on the "potential" for adverse healtheffects from any fibrous material of dimensions and aerodynamic diameter that will result in its beingrespirable. This is discussed in my response to Question 6(a) above. The toxicity of the substitutefibres and whether that information was determined on the basis of epidemiological or toxicologicalevidence is discussed in Question 6(c). The role of the chemical properties of fibres to induce cancer

WT/DS135/RPage 333

is not clear to me. The nature of the production process makes the substitute fibres more amenable tocontrol than asbestos fibres.

Dr. Musk:

5.395 I do not know: but it is my broad understanding that the physical and chemical properties ofthe substitute fibres suggests less risk of disease.

6.(c) The parties focus part of their arguments on cellulose fibres, para-aramid fibres, glassfibres and polyvinyl alcohol (PVA) fibres. What evidence exists with respect to the toxicity andhealth risks of these substitutes? Does the existing evidence suggest that these products areless/equally/more toxic than chrysotile asbestos fibres?

Dr. de Klerk:

5.396 See my response to Question 6(a).

Dr. Henderson:

5.397 In experimental studies on para-aramid fibres in comparison to chrysotile, Warheit et al. [12,257] found that p-aramid is bio-degradable in the lungs of exposed rats, with faster clearance thanlong chrysotile fibres which showed greater bio-persistence. In their 1996 study, these authors [12]found that:

" ... p-aramid is biodegradable in the lungs of exposed rats; in contrast, the clearance of long chrysotilefibres was slow or insignificant, resulting in a pulmonary retention of long chrysotile asbestos fibres.The dimensional changes of asbestos fibres as well as the pulmonary cell labelling data indicate thatchrysotile asbestos fibres may produce greater long-term pulmonary effects when compared to inhaledpara-aramid fibrils" [abstract].

5.398 The present status of knowledge has been summarized by Harrison et al. [19] in a recentreview of the comparative hazards of chrysotile and its substitutes:

"There are now practicable substitutes for the major remaining uses of chrysotile. Although lack of afull health and toxicological data set precludes a comprehensive assessment of the safety of substitutefibers, the application of basic principles of fiber toxicology enables a pragmatic decision to be madeon the relative safety of potential substitutes. Our judgement is based on relative considerations of theintrinsic properties of fibers, on the pathogenicity of chrysotile in comparison with that of substitutefibers, and on the potential for uncontrollable exposures. The three parameters of dose, dimension(especially diameter), and durability are key to determining the differential hazards. Due considerationof these factors leads us to the following conclusions regarding chrysotile and its main substitutes.

Chrysotile per se can cause lung cancer and asbestosis; it is less clear that chrysotile alone can causemesothelioma in humans, and indeed it may not, whereas tremolite and other amphiboles certainly cando so. There is no definitive evidence for a threshold exposure level for lung cancer induction,although some studies suggest that a threshold does exist.

The intrinsic hazardous properties of chrysotile can never be 'engineered out', and the potential for harmwill always remain. Prevention of ill health will thus always rely on the control of exposure, somethingthat history has shown cannot be guaranteed.

Unlike chrysotile, substitute fibres can often be designed or selected to have particular characteristics.Criteria for the substitution of asbestos by other fibers include a) the substitute fibers are not in therespirable range, do not readily fibrillate, and/or are less durable than chrysotile; b) other materials thatmust be incorporated into the replacement product do not, in combination with the replacement fiber,

WT/DS135/RPage 334

produce more harm overall than chrysotile alone; c) the replacement product has an equivalent oracceptable performance; and d) substitution would result in overall lower fiber exposures duringmanufacture and use and disposal, taking into account likely exposures. The same general principlecan be applied to substitute fibers others than those considered here.

We judge that PVA fibers will pose less risk than chrysotile because they are generally too large to berespirable, do not fibrillate, and the parent material causes little or no tissue reaction. Aramid fibershave a reduced potential for exposure when compared to chrysotile because they are generally of highdiameter and the production of respirable fibrils is energy intensive. The fibrils are less pathogenicthan chrysotile, are less biopersistent, and are biodegradable. Cellulose has the benefit of longexperience of use in a variety of industries without having raised significant concern. The potential forthe generation of respirable fibers seems to be less than is the case for chrysotile, although fibrillation ispossible. Cellulose is durable in the lung, and its biological properties should therefore be investigatedfurther. However, exposure levels for current uses are low, and it is biodegradable in the environment.

We believe that the continued use of chrysotile in asbestos-cement products is not justifiable in the faceof available and technically adequate substitutes. Likewise, there seems to be no justification for thecontinued residual use of chrysotile in friction materials" [pp 610-611].

5.399 From known past uses of asbestos and surveys of current uses in Australia, it is evident thatalternatives have replaced chrysotile to a large extent for the following products [NICNAS 99, p 111]:

"Products where chrysotile use has been completely replaced:

• Cement sheeting, tubes and piping.• Roofing tiles.• Textiles.• Fibre insulation.• Railway brake blocks.• Brake disc pads in new automotive vehicles (only 1 new vehicle model was identified

as being supplied with asbestos pads in Australia).

Products where a major proportion of chrysotile use has been replaced:

• Clutch facings (in automotive vehicles and industrial machinery e.g. tractors,centrifuge drives).

• Brake disc pads (in older taxi and courier vehicles, and industrial machinery).• Gaskets, such as spiral wound and head gaskets.• Washers.• Packing material.• Rotor blades (e.g. in high vacuum pumps)".

5.400 It is notable that chrysotile is no longer used for brake linings in new passenger cars producedin Australia by most manufacturers, having been replaced by substitute materials: NICNAS 99comments that:

"Out of 26 companies, 25 stated that they are using non-asbestos original equipment in all currentmodels. One company (Ford Motor Australia) reported that they are still using asbestos parts in twocurrent models: asbestos head gaskets for the Econovan and asbestos rear brake linings for the Fordutility. Ford Australia introduced non-asbestos components for their most popular models (e.g. Laser,Falcon and Fairlane) between 1989 and 1995. Other current models manufactured by Ford have beenasbestos-free since their introduction. ... Asbestos parts are imported by 6 of the 26 companies (BMW,Ford, Mazda, Mitsubishi, Nissan and Toyota) with five companies using asbestos parts for supersededvehicles and one company (Ford Australia) using asbestos parts in superseded and current models ... themajority of the vehicle manufacturing companies stated that they have had policies in place in regard tonot using asbestos components in new vehicles for the last 5 to 10 years" [p 22].

WT/DS135/RPage 335

5.401 This trend to use of brake linings free of asbestos is shown in the following Table 15 — incomparison to the usage of asbestos brake linings — between 1994 and 1998 (asbestos-containingbrake linings appear to be used primarily on older and superseded vehicle models).

TABLE 15: IMPORTS OF ASBESTOS AND NON-ASBESTOS BRAKE LININGS INTO AUSTRALIA, 1994-1998

Number of Articles

Import 1994 1995 1996 1997 1998 (Jan-Aug)

Asbestosbrake linings,passenger cars

492,295 47,735 43,087 771,182 (548,692)

Non-asbestosbrake linings,passenger cars

70,109 321,472 485,812 2,084,963 (4,057,143)

Source: NICNAS 99.

Dr. Infante:

5.402 There is no information to my knowledge that cellulose fibres, para-aramid fibres or ployvinylalcohol (PVA) fibres are carcinogenic. Cellulose fibres have not been studied experimentally forcarcinogenicity. It is noteworthy, however, that cellulose has been used in the paper industry forhundreds of years and to date an elevated risk of death from lung cancer and mesothelioma has notbeen observed. Excess incidences of pharyngeal and/or laryngeal cancers were reported in twostudies, but these observations have not been corroborated in other studies (IARC, 1987). Wood dustis associated with sino-nasal cancer, but not with lung cancer or mesothelioma. A relatively greaterrisk appears to be associated with hard woods as compared to soft woods, which suggests that thecellulose may not be the primary factor in the induction of these cancers. Workers exposed to cottondust also do not demonstrate an excess of lung cancer or mesothelioma even though they developbyssinosis. The debate as to whether this disease is due to cotton dust per se, or to contaminants of thecotton fibre, however, is not resolved.

5.403 Para-aramid fibrils have been studied for carcinogenicity in experimental animals byinhalation and by intra-peritoneal injection. No cancer response was observed. As concluded byIARC (1997), there is inadequate evidence for the carcinogenicity of para-aramid fibrils inexperimental animals. The carcinogenicity of para-aramid fibrils has not been evaluated in humans.Likewise, IARC (1987) concluded on the basis of its review of animal cancer tests that there is noevidence for the carcinogenicity of PVA fibres. PVA fibres have not been evaluated forcarcinogenicity in humans. It is also my opinion that there is no evidence that these fibres present anyrisk of cancer to humans.

5.404 With regard to glass fibres, IARC (1988) concluded there was sufficient evidence for thecarcinogenicity of glass wool in experimental animals and that there was inadequate evidence for thecarcinogenicity of glass wool to humans. Subsequent to the IARC (1988) review, my colleagues and Ihave reviewed the toxicological and epidemiological studies related to exposure to glass fibres. In ouropinion, there is conclusive evidence from implantation and inhalation studies that glass fibres arecarcinogenic in experimental animals (Infante et al. 1994). Studies of workers exposed to glass fibresalso demonstrate a significantly elevated risk of death from lung cancer. It is our interpretation ofthese studies that employment in the manufacturing of glass fibres carries with it an elevated risk ofdeath from lung cancer. Is it proven beyond a doubt through epidemiological study that fibrous glassis a human carcinogen? In my opinion, it is not. However, given the positive animal cancer testresults, knowledge that these fibres can be inhaled and retained in the lungs, evidence that workers

WT/DS135/RPage 336

employed in the manufacturing of these fibres die at a significantly elevated rate of lung cancer, it ismy opinion that it is more likely than not that glass fibres are carcinogenic to humans and thatemployment in this industry carries with it an elevated risk of death from lung cancer.

5.405 It is also my opinion that glass fibres are not as potent as chrysotile asbestos in causingdisease. With regard to the capability of glass fibres to cause lung cancer, I have previously publishedthe opinion that on a fibre-per-fibre basis, glass fibres may be as potent or even more potent thanasbestos in causing lung cancer. This opinion was based on epidemiological studies which generallydemonstrated 10 per cent to 20 per cent elevation in the relative risk of lung cancer (the 1987 study ofCanadian workers by Shannon demonstrated a 2-fold risk) as a result of exposures to glass fibres thatwere reported to be fairly low. Within the past year, however, I have had the opportunity to discussoccupational exposures during glass fibre manufacturing with workers formerly employed at theCanadian facility that manufactured fibrous glass. According to several workers, they were alsoexposed to crystalline silica many times over the permissible limit through the dumping of sand intothe hopper that was used to feed the furnace for melt down. They were exposed to asbestos that linedthe furnace when they removed the insulation from the oven doors by hand, or chiselled it away in theabsence of respiratory protection; they would then add water to asbestos fibre to make "asbestos mud"that was applied to the oven doors by hand, or by trowel. Workers were so uninformed of the hazardthat they sometimes would throw "asbestos mud balls" at each other. The workers at this facility werealso exposed to phenol formaldehyde resin that was used as a binder for the glass fibres; they werealso exposed to tar that was applied to paper that was then applied to the glass fibre pack. Exposuresto glass fibres only is mentioned in the study of these workers that was published by Shannon (1987).

5.406 Furthermore, there is less evidence from epidemiological studies that exposure to glass fibresis associated with a pneumoconiosis as compared to the data for chrysotile asbestos exposure andasbestosis. There is no evidence that exposure to glass fibres is associated with mesothelioma. Forthe few cases of mesothelioma that have been identified among workers exposed to glass fibres todate, there is claim that they also had been exposed to asbestos fibres. Therefore, it is difficult toattribute these cases of mesothelioma to the glass fibre exposures. Therefore, the totality of diseaserelated to chrysotile asbestos exposure would be greater than that related to a similar amount ofexposure to glass fibres.

5.407 In conclusion, only one of the fibres (glass fibres) that may play any significant role insubstitution for chrysotile asbestos demonstrates evidence of being carcinogenic. Data for the totaltoxicity related to these fibres, however, is less than that for chrysotile asbestos fibres. Cellulosefibres have not been tested for carcinogenicity in experimental animals, but epidemiological studies ofworkers exposed to cellulose in three separate industries, i.e., furniture manufacturing, cotton textilemanufacturing and the paper products industry, have not demonstrated an elevated risk of contractinglung cancer or mesothelioma. Regarding non-malignant lung disease among cotton dust exposedworkers, it is not known whether the cotton dust per se, or contaminants of the cotton fibre, areresponsible for the byssinosis observed in these workers.

Dr. Musk:

5.408 I understand that substitutes have been shown to be less toxic in animals.

3. Summary Comments by Dr. Henderson

5.409 In-place asbestos is widely distributed in industrialized societies and much includes mixturesof chrysotile and amphiboles — although chrysotile has been the predominant type of asbestos usedthroughout Western Europe for many years (about 94-97%).

WT/DS135/RPage 337

5.410 Lung cancer and mesothelioma are the most important bio-hazards from asbestos in place andthe continued use of asbestos.

5.411 Because of the prolonged lag-time between exposure and the subsequent development ofeither lung cancer or mesothelioma, most mesotheliomas in the 1990s and beyond can be attributed toexposures sustained decades before; the mesothelioma "epidemic" predicted for Europe over the nextthree decades can be attributed to exposures before, during and after the 1960s and 1970s, especiallyto one or more of the amphibole varieties.

5.412 For the amphibole forms of asbestos and mixtures of asbestos types, a linear dose-responserelationship has been found at high levels of exposure; a dose-response relationship with an increaseof the relative risk of mesothelioma to > 2.0 has also been observed at low levels of exposure, in theorder of 0.5-1.0 fibre-year (which overlaps with non-occupational environmental exposures). Nolower threshold dose for mesothelioma induction has been delineated for the amphiboles.

5.413 Chrysotile also has the capacity to induce mesothelioma, although it is lessmesotheliomagenic than the amphiboles (my estimate is 1/10th-1/30th).

5.414 Commercial Canadian chrysotile on average contains trace quantities of tremolite, includingfibrous tremolite (< 1%).

5.415 Tremolite — a non-commercial amphibole — also has the capacity to induce mesothelioma.

5.416 The carcinogenicity of Canadian chrysotile may be attributable to the trace tremolite content,but it is not possible to separate the dose-response effects for the chrysotile and the tremolite.

5.417 At high levels of exposure to Canadian chrysotile, a linear dose-response relationship hasbeen observed.

5.418 To the best of my knowledge, there are no epidemiological or observational data on dose-response effects of chrysotile only at low levels of exposure.

5.419 No lower threshold dose for the carcinogenic effects of chrysotile has been identified (EHC203).

5.420 To the best of my knowledge, there are no observational data on the potential carcinogeniceffects of inhaled chrysotile when superimposed upon a pre-existing burden of amphiboles ±chrysotile in lung tissue.

5.421 Although the amphiboles are far more potent than chrysotile for mesothelioma induction, thisdifferential in carcinogenicity may be less obvious or absent for lung cancer induction, but this is stillthe subject of some dispute; chrysotile is associated with a low risk of lung cancer among Canadianchrysotile miners and millers, but the highest risk for lung cancer induction has been observed forSouth Carolina asbestos textile workers who used Canadian chrysotile almost exclusively.

5.422 A linear dose-response relationship has also been observed for the risk of lung cancer versuscumulative asbestos exposure. Although some authorities favour a linear no-threshold model for lungcancer induction, others suggest that a threshold may exist, but this has not been delineated innumerical terms.

5.423 In contrast, asbestosis is a dose-dependent non-cancerous disorder, with clear evidence of athreshold effect, although the threshold may be lower than previously supposed, at least for

WT/DS135/RPage 338

histological asbestosis; there is no risk of asbestosis at low levels of chrysotile exposure.

5.424 Although reduction of airborne asbestos fibre concentrations in the mining and manufacturingindustries has been achieved, it is too early to evaluate the effects of these reduced exposures, becauseno epidemiological data are available; however, with reduction of cumulative exposures, a reductionin the incidence of both asbestos-related mesothelioma and asbestos-related lung cancer can beexpected.

5.425 The risks from low-level occupational exposure to chrysotile, or from occasional peakconcentrations, have not been delineated but are predictably small.

5.426 Carcinogenic hazards from ultra-low levels of atmospheric chrysotile fibres (e.g. simpleoccupancy of public buildings) appear to be minuscule, negligible or undetectable.

5.427 Therefore, health concerns over chrysotile dust exposure narrow down to a workplace issue.

5.428 There is evidence of an increased incidence of mesothelioma among, say, brake mechanics inAustralia exposed to chrysotile derived from brake blocks and lining.

5.429 With the reductions of airborne fibre concentrations in the asbestos mining, milling andmanufacturing industries, construction trades workers constitute the group of workers at greatest riskfrom exposure to asbestos-cement products (e.g. builders, builders' labourers, carpenters, electricians,plumbers and roofing workers). This group constitutes a large, disparate and non-cohesive workforcefor which controlled use of asbestos is not achievable, for the reasons discussed earlier in this report.

5.430 Therefore, chrysotile asbestos should not be used in building materials, because of the hazardsimposed by installation, maintenance and removal operations (EHC 203); these risks may becompounded for some groups by catastrophic events affecting buildings — e.g. fires (with a burst ofasbestos fibres into the atmosphere and the necessity for clean-up operations), and other disasters.

5.431 Substitutes for chrysotile are available for many applications (e.g. cellulose fibres, para-aramid fibres and polyvinyl alcohol); evidence indicates that these fibres are less bio-persistent thanchrysotile and, therefore, national health authorities (EHC 203, NICNAS 99) have recommendedphasing out or prohibition of chrysotile whenever safer substitute materials are available.

5.432 Therefore, from a perspective of caution and prudence for occupational health and safety, itfollows that chrysotile should either:

(a) Be restricted to only a few and well-defined applications so that it is inaccessible tothe great majority of workers and is available for use by only small and cohesivespecialized worker groups that can be trained effectively in its controlled use(e.g. analogous to nuclear fuels); in effect, this means that chrysotile should not beused in building products (e.g. high-density fibro-cement materials such as asbestos-cement sheets) or friction products.

OR

(b) It should be made inaccessible to everyone, by prohibition, unless the alternativespose equal or greater hazards and equal or greater problems with control.

5.433 These views are also expressed in EHC 203, wherein it is stated:

WT/DS135/RPage 339

"a) Exposure to chrysotile asbestos poses increased risks for asbestosis, lung cancer andmesothelioma in a dose-dependent manner. No threshold has been identified for carcinogenic risks.b) Where safer substitute materials for chrysotile are available, they should be considered for use.c) Some asbestos-containing products pose particular concern and chrysotile use in thesecircumstances is not recommended These uses include friable products with high exposure potential.Construction materials are of particular concern for several reasons. The construction industryworkforce is large and measures to control asbestos are difficult to institute. In-place building materialsmay also pose risk to those carrying out alterations, maintenance and demolition" … [p 144]d) The combined effects of chrysotile and other insoluble respirable particles needs further study.e) More epidemiological data are needed concerning cancer risks for populations exposed to fibrelevels below 1 f/ml, as well as continued surveillance of asbestos-populations" … [p 145]

5.434 NICNAS 99 sets out a similar set of recommendations:

• "Chrysotile is a known human carcinogen.

• Prudent OHS [occupational health & safety] policy and public health policy favoursthe elimination of chrysotile wherever possible and practicable.

• The main exposure to Australian workers arises from manufacture, processing andremoval of friction products and gaskets. Home mechanics are also exposed during'do-it-yourself' replacement of brake pads/shoes. ... . In Australia, chrysotile is nolonger used in high density materials such as chrysotile-cement.

• Current overseas experience with the phasing out of chrysotile products indicates thata range of alternatives is available to suit the majority of uses. Good OHS practicedictates that use of chrysotile products should be restricted to those uses wheresuitable substitutes are not available, and alternatives should continue to be sought forremaining uses".

5.435 Whether the objective of removal of chrysotile from the workplace and the generalenvironment is achievable by enforcement of controlled use for a few restricted applications — or byprohibition — is essentially a societal question and a public health policy issue. For the reasonsdiscussed in this report, a complete ban is more certain to accomplish this objective(paragraph 5.432(b)). Therefore, as a cautious and prudent approach to national occupational healthpolicy, a complete ban is neither unreasoned nor unreasonable; on the balance of prevailing scientificevidence and uncertainties discussed in this report, such a policy seems defensible and, arguably,justifiable as a national health measure. Perhaps it is best to let Bradford Hill have the last words:

"All scientific work is incomplete — whether it be observational or experimental. All scientific work isliable to be upset or modified by advancing knowledge. That does not confer upon us a freedom toignore the knowledge we already have, or to postpone action that it appears to demand at a given time."

4. Endnote by Dr. Henderson

5.436 Wishing to add two further pertinent references after completing his Report, Dr. Hendersonattached the following Endnote. These references27 deal with the following:

5.437 Clearance of chrysotile fibres from human lung tissue: In the past, the kinetics of chrysotileclearance from lung tissue have been investigated mainly in experimental models using rodents. In anautopsy study published in 1999, Finkelstein and Dufresne [1] investigated clearance of chrysotile

27For complete references, see Annex III to this Panel Report.

WT/DS135/RPage 340

from the lung tissue of 72 Quebec chrysotile miners and millers in comparison to 49 control subjects,using regression analyses, with the following findings:

• There was a significant association between the duration of occupational exposureand the tissue burdens of chrysotile and tremolite.

• The concentration of chrysotile decreased with time after exposure ceased but theconcentration of tremolite did not.

• The clearance rate varied inversely with the length of chrysotile fibres. For fibres> 10 µm in length - i.e. fibre lengths in the reported range for carcinogenicity - theclearance half-time was estimated to be eight years. In other words, the tissuebio-persistence of chrysotile fibres in this study seems substantially more prolongedthan in rodent experiments, and presumably corresponds to persistent high chrysotilefibre concentrations for many years after cessation of occupational exposure inhumans, as discussed in paragraphs 5.112 - 5.113. It is also notable that theconcentration of 6,250,000 chrysotile fibres mentioned in those paragraphs (for anindividual but by no means unusual patient) is probably above the level at whichRogers et al. [2] identified an odds ratio for mesothelioma of > 8.5 (even allowing fordifferences in fibre size between the two different laboratories), and even the durationof 16 years after exposure stopped (as opposed to its commencement: 24 years) fallsinto the lag-time range lung cancer induction by asbestos.

• Studies like this suggest that clearance mechanisms can be overwhelmed and breakdown at occupational levels of exposure in humans, with the existence of a long-termsequestered fraction of chrysotile fibres.

5.438 Mesothelioma rates in men and women in Sweden: attached to this Endnote is a recent paperby Jarvholm et al. [3] on trends in mesothelioma incidence in Sweden, which re-emphasizes some ofthe points made earlier in this report.

D. COMMENTS BY THE PARTIES ON THE RESPONSES FROM THE EXPERTS

1. Canada

5.439 Canada is pleased that the experts agree with Canada on certain crucial aspects of the debatein this case. Most importantly they opine that:

• Chrysotile is significantly safer than amphibole asbestos (three of the four expertsagree);

• there is no risk to the public from low-level environmental exposure to chrysotile orfrom exposure in buildings that contain chrysotile (all of the experts agree);

• there is no risk to workers in mines or factories where use of chrysotile is controlled(three of the four experts agree); and

• there is no risk to "handymen" or "do-it-yourselfers" who disturb chrysotile products,because their exposure is intermittent and thus inconsequential (three of the fourexperts appear to agree).

5.440 In short, although the experts agree on the inadequacy of a data (statistical limitation tosupport a threshold), their findings are consistent with the view that low levels of exposure to

WT/DS135/RPage 341

chrysotile asbestos create no detectable health risks. Indeed, the only population that the experts viewas having problematic exposure is tradesmen, e.g., plumbers, electricians and mechanics, who disturbor modify chrysotile cement and friction products. At this point, the experts and Canada diverge; also,at this point, the experts stray beyond their specialities (as several admit). Canada maintains thatadequate controls for these exposures can be developed and applied and has set forth such controls inComments to Experts' Answers to Question 5.

5.441 Several other aspects of the experts' answers require comment. Some of the responses of theexperts appear not to distinguish between chrysotile and amphibole exposure, and between modernuses (e.g., chrysotile friction and cement products) and historical uses (e.g., insulation containingamphiboles). In many places, for example, the experts appear to draw conclusions regardingchrysotile based in part or in whole on data from individuals exposed to amphiboles and/oramphiboles and chrysotile. This is of greatest concern to Canada regarding the experts' conclusionson tradesmen; as the experts no doubt agree, the greatest risk to tradesmen is not exposure to modernchrysotile products, but the disturbance of flocking or insulation containing amphiboles. Similarly,the experts do not always distinguish peak and cumulative exposures. For the purposes of defininghealth risks, the cumulative measure, not the peak measure, is key.

5.442 It is crucial that this proceeding forms on the pertinent issues. The key issue is whetherexposure to modern uses of chrysotile can be controlled to ensure worker safety or if a total ban isrequired to achieve an equivalent level of safety. The experts' answers help to focus the proceeding ontradesmen exposure.

Question 1(a)

5.443 Canada believes that the Panel should take note of the clarifications to this question proposedby Drs. de Klerk and Musk. The former writes that "the more relevant question here is: who is likelyto receive the most exposure and therefore have the greatest risk of disease". Dr. Musk rephrases thequestion in almost identical terms, taking the expression "risk of exposure" to mean "who is mostlikely to receive the most exposure and therefore be at the greatest risk of developing asbestos-relateddisease". In Canada's view, because the evidence suggests different risk per unit fibre exposure indifferent sectors, it is the combination of level of exposure, duration of exposure and risk per unit fibreexposure that is important.

5.444 The experts have confirmed that any risk from exposure to chrysotile will depend on thenature of an individual's specific occupational setting and the risk per unit fibre exposure in thatsetting, certain sectors being the subject of more stringent controls than others. For example, theexperts echo the Parties' agreement to the effect that the mining and manufacturing sectors havesuccessfully controlled the risks to which their workers had previously been exposed. Certain settingspose lower risk per unit fibre exposure than others do.

5.445 Canada does not disagree with the statement that the so-called secondary user sector is themost diverse. Canada nevertheless understands that the experts do not believe that the diversity of thisparticular workforce is to be considered the only factor that may contribute to a greater likelihood ofexposure; rather, as Dr. Musk puts it, "the risk of developing asbestos-related disease [...] (also)depend(s) on [...] the type of asbestos being produced or used or otherwise encountered. It wouldalso depend on the conditions of work such as indoors versus outdoor etc." As the Panel also knows,the specific uses or products also entail more or less risk.

5.446 Canada does not believe that the diversity of this workforce precludes effective control. Thediversity of a specific workforce is not indicative of the quality of the work practices actuallyobserved by the members of that workforce. A typical construction site offers numerous examples ofsound safety practices: from hard hats to proper footwear, from the use of common sense to following

WT/DS135/RPage 342

trade-specific work practises, measures are taken to insure safety and avoid trauma.

5.447 Canada notes that the experts have not commented on the assertion by theEuropean Communities that there is a correlation between the amount of chrysotile used by Franceand the incidence of asbestos-related disease. Clearly no such correlation can be made, in logic or infact. The logic on which the European Communities purports to base this assertion is a sophism, andshould be dismissed accordingly. From a factual point of view, the following factors suggest that thecorrelation is false: the relative difference in potency and in biopersistence of amphiboles andchrysotile, the historical uses of each fibre type, and the differences in risk per unit fibre exposure indifferent sectors.

5.448 Canada notes that Dr. Infante assimilates friable amphibole or mixed fibre type exposurecircumstances to those of high-density chrysotile products, thereby answering the wrong question. Hecorrectly identifies worker contact with insulation as being the "typical scenario" in which exposure toasbestos will occur. But most insulation is friable, as opposed to high-density, and most friableinsulation products contained amphiboles or mixed fibre types. It is not clear how this answer basedon friable mixed asbestos products responds to a question relating solely to safety of high-densitychrysotile products.

Question 1(b)

5.449 Canada takes note that the experts have indicated that the risk of human health associated withthe various uses of chrysotile throughout its life cycle is overwhelmingly a workplace issue, andtherefore not related to the "handyman".28

Question 1(c)

5.450 The answers given by the experts indicate that, on their own, chrysotile cement products donot pose a health risk because of their normal weathering, erosion or general degradation, and that"there is little or no dispute among experts on this issue".29

5.451 Canada wishes to draw the attention of the Panel to the results of the investigation carried outby the Western Australia Advisory Committee on Hazardous Substances (WAACHS), cited byDr. Henderson.30 This report contains different sections describing asbestos cement products, theirproduction and use and their health effects, as well as surveys of schools and other relevantmeasurements of asbestos concentrations. In addition to pertinent recommendations, the reportcontains several appendices, including one on the Effects of Asbestos Cement Products – A Review ofthe Literature and another on Acceptable Air Concentrations of Asbestos Fibres in the GeneralEnvironment, both prepared by one of the experts to this Panel, Dr. de Klerk.

5.452 On low level air concentrations, Dr. de Klerk writes: "[M]ost of these estimates are on orbelow the level of what the Royal Society would consider acceptable [...] The 1986 IPCS report didnot even bother to estimate such risks and summarised the risk exposure unrelated to occupation asbeing undetectably low".31 Indeed, the executive summary of the WAACHS report indicates: "[...][T]he level of risk is low enough to be considered to be negligible relative to these other risks in our

28Canada notes that, according to Dr. Henderson: "[f]rom my perspective, this is overwhelmingly aworkplace issue […]".

29Henderson, answer to Question 1(d).30Henderson, p. 54, citing Multiple Authors, Asbestos Cement Products. Report by the

Western Australia Advisory Committee on Hazardous Substances, Perth, 1990, hereinafter the WAACHSReport.

31De Klerk, N., Acceptable Air Concentrations of Asbestos Fibres in the General Environment. AReview of Scientific Evidence and Opinion in the WAACHS Report, Appendix 3, p. 10.

WT/DS135/RPage 343

society".32 Similarly, in his report to the Panel, Dr. Henderson underlines that compared to the fibreconcentrations observed in the vicinity of asbestos-cement roofing, "a greater risk to health wouldarise from [workers] falling from or through the roofs".33

5.453 High-density chrysotile on buildings has been extensively studied. Indeed Teichert found thefollowing: "the study of emission conducted on coated and uncoated roofing materials revealed lowasbestos fibre concentrations, even though severe corrosion was observed on uncoated asbestoscement roofs and a considerable quantity of material containing asbestos could be removed byblowing or suction. The asbestos fibre concentrations that were measured in populated areas are wellbelow the level considered acceptable by the Health Authorities of the Federal Republic of Germany,i.e. clearly below 1000 fibres/m3 (or 0.001 f/ml)".34 Felbermayer and Ussar, for their part, write: "acomparison of the asbestos fibre concentrations in those areas with and without asbestos-cementroofing (...) lead to the conclusion that there is no statistical significant connection between the use ofasbestos-cement materials and the asbestos fibre concentrations found in the various measurementareas."35

5.454 Finally, Canada would like to bring to the Panel's attention the following recommendation ofthe WAACHS report, which is: "[A]n asbestos cement roof, which has not deteriorated to an extentwhere physical safety or structural integrity is of concern, should not be replaced. In addition, anasbestos cement roof should not be treated with a coating on the basis of risk to health. Other asbestoscement products are generally less prone to deterioration and do not require attention for healthpurposes".36 Nonetheless, many chrysotile-cement products are coated with protective sealant agents.

Question 1(d)

5.455 The experts agree that the degree of risk to the health of workers intervening on high-densitychrysotile cement products will depend on the manner in which an intervention is carried out. Asnoted by Dr. Infante in his response to question 1(e), "the extent of the exposure to the worker (...)would depend on the nature of the intervention, e.g., the circumstances under which the chrysotileasbestos product is manipulated in terms of work practices, the controls, or lack of controls in placeand the type of personal protective equipment provided to the worker". Dr. Henderson illustrates thisproposition when he writes that "cutting (chrysotile-cement) with hand saws produced lowerconcentrations."

5.456 Canada accepts that abrasion and cutting of high-density chrysotile products can releasematerials. However, the degree of exposure, if any, will depend on the methods and controls used.Canada notes that the experts disagree as to the exact composition of the materials that would bereleased by such interventions (see question 1 (f)), although there is apparent agreement that cuttingchrysotile cement releases crystalline silica, an IARC Class 1 carcinogen.37 Cutting chrysotile cementusing simple work practices such as those outlined in ISO Standard 7337 will therefore provideprotection from any potentially harmful material contained in such a product. Wetting the productbefore cutting and/or using commonly found suction attachments when sawing are techniques that canbe used as added, but perhaps unnecessary, precautions. A final safety barrier would be for theworker to wear a facemask: this step would render it virtually impossible for the worker to inhale dust.

32WAACHS Report, p. 2.33Henderson, answer to Question 1(b).34Teichert, U., Immissionen durch Asbestzement-Produkte, (1986) Teil 1 Stub Reinhaltung der Luft,

Vol. 46, No. 10, pp. 432-434.35Felbermayer, W., Ussar, M. B., Research Report: Airborne Asbestos Fibres Eroded from Asbestos

Cement Sheets, (1980) Institut für Umweltschutz und Emissionesfragen, Leoben, Austria.36WAACHS Report, p. 4.37Infante, answer to Question 1(f).

WT/DS135/RPage 344

5.457 Neither the European Communities nor the experts have demonstrated that such practiceswould subject workers to cumulative exposures presenting health risks. An American surveyestimated that a worker would spend less than 1/16th of his work time on tasks that would involveaggressive interventions on chrysotile-cement of the type susceptible of releasing any substantialamount of dust.38 Canada submits that the European Communities have not identified any populationof workers that would be subject to a detectable risk because of professional contact with high-densitychrysotile cement. The European Communities' contentions vis-à-vis the "handyman" are thereforeeven less convincing (see next answer).

Question 1(e)

5.458 Canada agrees with Dr. Henderson's conclusion that "occasional interventions (...) wouldpredictably produce low cumulative exposures, with a lower risk (...)". Dr. Henderson also affirms thatfor "electricians, carpenters, plumbers, insulation workers and so forth", "it is acknowledged that mostif not all these mesotheliomas are a consequence of exposure to (...) a mixture of asbestos types,including chrysotile and one or more of the amphiboles."

5.459 The Panel has not been presented with evidence that contradicts Canada's assertion thatoccasional interventions do not pose a risk that is significantly different from zero (statistically).Therefore, the experts have not validated the EC's claim that an alleged risk for workers or the"handyman" is something more than undetectable.

5.460 Nor has the Panel been presented with evidence or expert opinion that supports theEuropean Communities' claim with respect to the "handyman". Given that cohorts exposed torelatively high concentrations of chrysotile over entire occupational lifetimes show no increase ofdisease, it is unlikely that occasional interventions by a "handyman" would produce more than anequally undetectable risk. Obviously the "handyman" or bricoleur du dimanche will not encounterhigh-density chrysotile-cement products on a daily basis, nor devote his "handiwork" exclusively orprincipally to cutting such products. Rather, the typical "handyman" will rarely, if ever, come intocontact with chrysotile cement products, let alone be sawing them.

5.461 The Panel should note that no evidence has been presented that shows any fatality in workers,let alone in "handymen", who would have been subject to any form of exposure, high or low, fromcontact with chrysotile cement products; the argument presented by the European Communities hasbeen based entirely on hypothetical scenarios.39

Question 1(f)

5.462 There is debate in the scientific community and among the experts appointed by the Panel asto the exact physical and chemical composition of what is contained in dust from certain interventionson chrysotile cement products. Dr. Infante writes, however, that this dust (indeed, all cement dust)will contain "crystalline silica", a known IARC Class 1 carcinogen found in all cement.

5.463 A 1992 IARC publication determined that "in asbestos-cement products, the asbestos fibresusually represent 10-15 per cent of the total weight and are embedded in the cement. Therefore, it isnot certain a priori that dust generated from asbestos-cement products will have the same effect asdust from pure chrysotile. [I]n asbestos-cement dust most of the asbestos fibres form aggregates withcement particles ... [t]hose which do not form aggregates ... appear to be coated with a calcium-containing layer. In absorption experiments, the asbestos-cement dust behaves more like cement dust

38CONSAD Research Corporation, 1990, No. 8282.39See Canada's Comments to Questions 5(c) and (e).

WT/DS135/RPage 345

than like asbestos dust."40 Because the surface properties of asbestos fibres are altered by certainheating, pH, and abrasion conditions41, it can be deduced that the composition and effect of the finalaerosol would be different than that suggested by studies of concentrations of fibres alone. And again,controlled use procedures limit release, and proper breathing equipment precludes exposure.

Question 1(g)

5.464 Canada believes that the Panel was not presented with any quantification of this risk, orindeed its existence. Dr. Infante describes how the removal of chrysotile cement panels can beaccomplished with negligible release of respirable fibres. Most other chrysotile cement products arefound in the form of underground water pipes. Studies show that these products remain intact fordecades after installation.42 Hence, very little of this product will need to be disturbed. Moreover, theexcavation and removal of pipes is not executed by manual labour, the bulk of any removal beingdone by heavy machinery.

5.465 The Panel should also note that the removal of chrysotile cement products does not generallyentail crushing. Rather, if and when necessary, chrysotile cement products can be removed,transported, and disposed of by means that do not constitute a detectable risk to human health. TheFrench Circulaire 97-15 accomplishes this goal for the high-density products at issue in thisproceeding.43 Also, if France is ensuring the safe removal and disposal of friable asbestos materials44

known to contain amphiboles or mixed fibre types, the Panel should conclude that the removal anddisposal of high-density chrysotile cement products can be accomplished even more safely,since high-density materials are indisputably recognised, even by France, as much easier to managethan anything in friable form.45

40Characterization and Proprieties of Asbestos-Cement Dust, in Biological Effects of Mineral Fibres,

Vol. 1, IARC Scientific Publications No. 30, Lyon, 1980, pp. 43, 49 and 50.41Henderson, answer to Question 2.42Canada notes that a study of 15 water supply systems in the State of Illinois U.S.A., where some

asbestos-cement pipes were up to 40 years old, and where the water was non-aggressive to moderatelyaggressive, shows no significant differences in the water before and after passing through the asbestos-cementpipe network: Hallenbeck, W. H., et al., Is Chrysotile Asbestos Released from Asbestos Cement Pipe intoDrinking Water, (1978) Journal of the American Water Works Association 70 (2): 97-102.

43Circulaire no. 97-15 du 9 janvier 1997 relative à l'élimination des déchets d'amiante-ciment généréslors des travaux de réhabilitation et de démolition du bâtiment et des travaux publics, des produits d'amiante-ciment retirés de la vente et provenant des industries de fabrication d'amiante-ciment et des points de vente ainsique tous autres stock.

44Circulaire no. 96-60 du 19 juillet 1996 relative à l'élimination des déchets générés lors des travauxrelatifs aux flocages et aux calorifugeages contenant de l'amiante dans le bâtiment.

45Canada notes that French regulations indeed recognize a difference in disposal proscriptions between"les matériaux friables" and "l'amiante liée" – see Note DPPR/SDPD/BGTD/LT/LT no. 97-320 du 12 mars 1997relative aux conséquences de l'interdiction de l'amiante et à l'élimination des déchets, which is as follows:

"III. – Quelles sont les fillières d'élimination des déchets contenant de l'amiante?"Deux circulaires ont été diffusées, l'une le 19 juillet 1996 pour les déchets issus des travaux relatifsaux flocages et aux calorifugeages, l'autre le 9 janvier 1997 pour les déchets d'amiante-ciment."Les fillières d'élimination des déchets contenant de l'amiante autres que ceux qui ont fait l'objet desdeux cricularies précitées peuvent être déterminées par analogie aux prescriptions de ces deuxcirculaires:- Les matériaux friables, c'est-à-dire les matériaux susceptibles d'émettre des fibres sous l'effet

de chocs, de vibrations ou de mouvements d'air, sont assimilables aux flocages et auxcalorifugeages. Ils devront être éliminés dans des installations de stockage des déchetsindustriels spéciaux ou dans l'unité de vitrification;

- pour les déchets contenant de l'amiante liée, trois cas sont envisageables:

WT/DS135/RPage 346

Question 1(h)

5.466 See comments on previous question.

Question 1(i)

5.467 Canada wishes to add the following comments on the answers to this question. Once removedfrom a building, a chrysotile cement panel, even if broken into several pieces, remains as intact aswhen it formed part of that building. Studies referred to above indicate that chrysotile cement roofingdoes not contribute (< 0.001 f/ml) to the levels of chrysotile occurring naturally in the environment.Likewise, chrysotile cement piping is generally found below ground, and therefore does not contributeto the levels of chrysotile naturally occurring in the atmosphere. If removed from roofing, or ifexcavated and removed from a water system, chrysotile cement products are transported to a landfilland buried anew beneath a layer of earth. Consequently, Canada is of the view that used chrysotilecement products can be eliminated safely.

5.468 Canada also notes that recent technology has enabled safe (in some cases, on-site) disposal ofchrysotile products. For example, chrysotile can be treated with chemicals and/or subjected to hightemperatures so as to render the end product entirely harmless and, in fact, suitable for enhancing thequality of soils. For example, in the United States, a foam has been developed that eliminates the riskassociated with removing asbestos from buildings; when this product is sprayed onto asbestosfireproofing, the fibres turn into harmless globs of magnesium silica. A U.S. building contractorrecycles asbestos by subjecting it to a chemical bath and high temperatures resulting in a totally inertend-product suitable for soil improvement. A Japanese company, responding to a government lawmandating pollution-free disposal of asbestos, melts asbestos into harmless glass.46

Question 2

5.469 Canada has advocated the use of chrysotile in high-density products only; textiles are not ofthat category, and had been banned in France prior to the adoption of the measure that is the subject ofthis dispute. Friction materials using chrysotile have not been shown to constitute a risk to humanhealth.47 Indeed, the contrary is probably true: lesser braking action of linings manufactured without

- Si les déchets sont composés d'amiante associée uniquement avec des matériaux

inertes, ceux-ci pourront être éliminés conformément à la circulaire du 9 janvier1997 relativement à l'élimination des déchets d'amiante-ciment;

- si l'amiante est associée avec des matériaux, qui lorsqu'ils deviennent des déchets,sont classés déchets ménagers et assimilés, c'est par exemple le cas des dalles vinyl-amiante, ils pourront être éliminés dans des installations de stockage de déchetsménagers et assimilés;

- si l'amiante est associée avec des matériaux, qui lorsqu'ils deviennent des déchets,sont classés déchets industriels spéciaux, ils devront être éliminés soit dans desinstallations de stockage de déchets industriels spéciaux, soit dans l'unité devitrification."Dans tous les cas, l'industriel ou l'entreprise devra fournir des éléments permettantde caractériser les déchets afin de déterminer les filières d'élimination adaptées."

46Acid v. Asbestos, Discover, Information Access Company, No. 7, Vol. 20, July 1, 1999, p. 102.;Contractor Recycles Asbestos for Re-Use in Construction, Air Conditioning, Heating & Refrigeration News,Business News Publishing Company, Vol. 194, No. 2, January 9, 1995, p.1; Kent Firm Fires up New Asbestos-Disposal System, Puget Sound Business Journal, Vol. 13, No. 14, August 21, 1992, p. 9; Japanese Plant TurnsAsbestos into Glass, American Metal Market, Vol. 100, No. 145, July 28, 1992, p. 4.

47Appendix A on Control Use in the Friction Industry, Canada's Comments to Question 5(a), containedin Annex IV to this Report.

WT/DS135/RPage 347

chrysotile is cited by France as the safety concern for which it exempted certain military vehicles fromthe purview of the Decree.48

Question 3(a)

5.470 Three of the four experts concur with the position of Canada and the WHO that a cleardistinction must be made between chrysotile and amphiboles. Dr. Musk believes that "there is a needto distinguish chrysotile asbestos from amphiboles based on the epidemiological data at least" and thatthe relative pathogenicity of some amphiboles to chrysotile may, in some cases such as mesothelioma,be 100 to 1. Dr. de Klerk affirms that the "epidemiological evidence is clear that, for a given quantity(intensity and duration) of exposure, chrysotile imparts less risk than amphibole fibres." Thedifference in pathogenicity is, according to Dr. de Klerk, up to 50-fold in the case of lung cancer andup to 100-fold for mesothelioma. Dr. Henderson concludes that: "a clear distinction should be madebetween chrysotile and the amphibole forms of asbestos."

5.471 Domestic legislation and international standards have long recognized the relativepathogenicity of different asbestos fibre types by permitting higher exposures to chrysotile than toamphiboles. In the European Communities in 1998, for example, the maximum exposure level foramphiboles was 0.3 f/ml, whereas it was 0.6 f/ml for chrysotile. In Canada (Quebec), it is 0.2 f/ml forcrocidolite and 1 f/ml for chrysotile. Similarly, international instruments such as the ILO'sConvention 162 and Recommendation 172 advocate an outright ban on crocidolite, whilerecommending replacing chrysotile if and only if safer substitutes exist.

5.472 Dr. Infante acknowledges epidemiological data to the effect that chrysotile is less dangerousthan amphiboles, but sees no basis for distinguishing between asbestos fibre types. Dr. Infante'sdissident view to the question of relative pathogenicity between asbestos fibres – one which echoesthe European Communities' argument but simply begs the question – is that because amphiboles andchrysotile are both classified as carcinogens, no distinction should be made.

5.473 In 1998, the WHO affirmed that a distinction should be made between chrysotile andamphiboles because using data from exposures to amphiboles "contribute[s] less to our understandingof the effects of chrysotile, due to concomitant exposure to amphiboles."49 The distinction betweenchrysotile and amphiboles is crucial in this instance since the current problem of asbestos in France isdue to past uses of friable materials, high-level exposures, and the use of amphibole fibres. Thedistinction between chrysotile and amphibole asbestos is also important because the extrapolationsmade by INSERM to assess the risks associated with chrysotile are based on exposures to amphibolefibres in proportions of up to 100 per cent in circumstances which have nothing to do with the currentuses of chrysotile.50

Question 3(b)

5.474 Physical properties, as well as chemical properties that determine biopersistence, areidentified as relevant factors of pathogenicity by Drs. Musk, Henderson and de Klerk and by theWHO.51

5.475 Dr. de Klerk, for example, has written that:

48Article Ier 2a) Arrêté du 17 mars 1998 relatif aux exceptions à l'interdiction de l'amiante.49WHO, IPCS Health Criteria 203 on Chrysotile, WHO, Geneva, 1998, p. 107.50See notably INSERM Report, p. 213.51WHO, IPCS Health Criteria 203 on Chrysotile, WHO, Geneva, 1998, p. 51: "[I]t is considered that

the potential respiratory health effects related to [...] airborne concentrations, patterns of exposure, fibre shape,diameter and length (which affect lung deposition and clearance) and biopersistence."

WT/DS135/RPage 348

"[T]he important carcinogenic properties of asbestos are related to the physical properties of size andshape of the fibers, and to their quantity. To cause any harm, fibers must be able to reach the targetorgans [...]." […]

"[I]n all occupationally exposed series of mesotheliomas, none have occurred in cohorts whereamphibole asbestos has never been used or detected. Chrysotile asbestos has not been directlyimplicated in any case of peritoneal mesothelioma. [...] The main differences between the effects ofchrysotile and amphibole fibers are:

1. Industries using a mixture of asbestos types have higher rates of disease than similar industriesusing only chrysotile.

2. Chrysotile fibers are eliminated more readily from the lungs than are amphibole fibers.

3. Much smaller doses of amphibole fibers than chrysotile fibers can induce mesothelioma."52

5.476 All four experts recognize the lower biopersistence of chrysotile. INSERM, citing numerousstudies, also acknowledges the lower biopersistence of chrysotile:

"Les études expérimentales ont montré que la biopersistance des fibres de chrysotile était inférieure àcelle des amphiboles (Wagner et al., 1974; Davis et al.; Davis and Jones, 1988, Churg et al., 1989;Churg, 1994)."53

5.477 Dr. Infante identifies the physical characteristics as also relevant to the relative pathogenicityof asbestos fibre types, but, unlike the three other experts and the WHO, believes that the role ofbiopersistence, through the element of solubility, "is not so clear."

5.478 Chrysotile fibres are "curly" and downy while amphibole fibres are straight and rigid likeneedles.54 Drs. de Klerk and Musk both specifically address the "straightness" element. The WHOhas observed that:

"Inhalation of respirable straight fibres [amphiboles] is reported to be associated with greaterpenetration to the terminal bronchioles than in the case of 'curly' fibres [chrysotile]."55

5.479 Once they have entered the respiratory tract, chrysotile asbestos fibres, because of their curlyshape, are more easily cleared by the mucociliary process than are straight and rigid amphibolefibres.56 Dr. Henderson writes: "[I]t is well known that chrysotile fibres are cleared more rapidly thanamphiboles, especially in long-term studies (Churg, 1994)."57 This is confirmed by a 1994 Europeanstudy by Dr. Albin: "[A]dverse effects are associated rather with the fibres retained (amphiboles),than with the ones being cleared (largely chrysotile)."58

5.480 For chrysotile fibres that do nonetheless manage to become lodged in the lungs, the solubilityof the fibres and the action of macrophages come into play to make chrysotile a much less potentfibre. First, as the WHO recognizes, chrysotile has a lower resistance than amphiboles in acidic

52de Klerk, N.H. and Armstrong, B.K., The Epidemiology of Asbestos and Mesothelioma in Malignant

Mesothelioma, Henderson, D.W. et al., eds. Hemisphere Publishing, New York, 1992, 223 at p. 230.53INSERM Report, p. 92.54WHO, IPCS Health Criteria 203 on Chrysotile, WHO, Geneva, 1998, p. 11.55WHO, IPCS Health Criteria 203 on Chrysotile, WHO, Geneva, 1998, p. 11.56Kumar, V., Cotran, R. and Robbins, S., Basic Pathology, 6th Ed., London, Saunders Co., 1997,

p. 228.57Henderson, see above para. 5.112.58Albin, M., et al., Retention Patterns of Asbestos Fibres in Lung Tissue Among Asbestos Cement

Workers (1994) 51 J. of Occupational Environmental Medicine 205.

WT/DS135/RPage 349

environments such as the lungs.59 Second, the macrophages responsible for eliminating fibres from thelungs are able to deal more easily with chrysotile fibres than with amphibole fibres. A 1997 report ofthe French Government (G2SAT) referred to by the European Communities, recognizes that as aresult of the chemical dissolution process that takes place in the lungs, carcinogen activity issubsequently practically nil:

"Il a été démontré que le chrysotile est nettement plus facilement éliminé du poumon humain que lesautres formes [amphiboles]. Par ailleurs, il ne présente pratiquement plus d'activité cancérogène (parinjection intra-cavitaire) après attaque acide, laquelle dissout la majorité du magnésium."60

5.481 Dr. Wagner, in his 1988 study of asbestos-related diseases, concluded:

"Chrysotile is the least harmful form of asbestos in every respect and [...] more emphasis should be laidon the different biological effects of amphibole and serpentine asbestos fibre."61

5.482 It should also be noted that gravimetric comparisons between amphiboles and chrysotile –widely used in the past in experimental work – tend to grossly misrepresent the relative pathogenicityof the fibres. According to the WHO, chrysotile "may contain more than 10 times more fibres per unitweight."62 Recent studies that use both the fibre mass and the number of fibres as dose units confirmthat, on a per fibre basis, amphiboles are far more pathogenic than chrysotile.63

Question 3(c)

(i) Asbestosis

5.483 Dr. Henderson asserts that: "[T]he amphibole varieties of asbestos appear to be substantiallymore pathogenic than chrysotile for the induction of asbestosis and mesothelioma." According toDr. Henderson, "[A]sbestosis is a dose-dependent disorder with a threshold effect [...] There iswidespread agreement that asbestosis in general is a consequence of high intensity exposure (or lowerintensity but more prolonged exposure)."

5.484 INSERM also supports the existence of a threshold for asbestosis,64 and according toINSERM, current low-level exposures to chrysotile pose no threat of asbestosis: "les expositionsactuellement relevées dans les industries directement utilisatrices d'amiante devraient conduire à ladisparition des cas d'asbestose confirmée (Doll et Peto, 1985)."65 It is clear, therefore, that asbestosisis not relevant to this dispute.

(ii) Lung Cancer

5.485 Dr. Musk believes that lung cancer risks are more than ten times greater in the case ofamphiboles than in the case of chrysotile asbestos. Dr. de Klerk suggests the difference may be upto 50-fold.

59WHO, IPCS Health Criteria 203 on Chrysotile, WHO, Geneva, 1998, p. 4. Kumar, V., Cotran, R.and Robbins, S., Basic Pathology, 6th ed., London, Saunders Co., 1997, pp. 227; INSERM Report, p. 396.

60INRS, Rapport du Groupe scientifique pour la surveillance des atmosphères de travail (G2SAT),1997, p. 47.

61Wagner, J.C. et al., Correlation between Fibre Content of the Lung and Disease in East LondonAsbestos Factory Workers, (1988) 45 British Journal of Industrial Medicine 305.

62WHO, IPCS Health Criteria 203 on Chrysotile, WHO, Geneva, 1998, p. 69.63See WHO, IPCS Health Criteria 203 on Chrysotile, WHO, Geneva, 1998, p. 69 and 81; INSERM

Report, Table 2, p. 196; EPA, Integrated Risk Information System, Asbestos, Document No. CASRN 1332-21-4 on-line: EPA, <http://www.epa.gov/ngispgm3/iris/subst/0371.htm> (access date: June 10, 1999).

64INSERM Report, p. 327.65INSERM Report, p. 327.

WT/DS135/RPage 350

5.486 Dr. Henderson states that the "greater carcinogenicity of the amphiboles [...] appears not toextend to the induction of lung cancer"66 but he admits that "chrysotile is implicated in one of thelowest rates of asbestos-associated lung cancer (in Quebec chrysotile miners and millers)."67

Dr. Henderson's reluctance to conclude the greater carcinogenicity of amphiboles seems to be causedby the results of Dr. Dement's study of the Charleston, South Carolina asbestos textile industry.68

5.487 The Charleston data has recently been revisited by Bruce Case, André Dufresne,A.D. McDonald, J.C. McDonald and Patrick Sébastien in a study released in Maastricht inOctober 1999 at the VIIth International Symposium on Inhaled Particles, a symposium attended bysome of the world's leading experts. This study shows that a significant amount of crocidolite andamosite fibres was found in the textile workers' lungs. This analysis sheds new light on the issue andexplains the extreme results of the original study by Dr. Dement69 and the subsequent study byDr. Stayner.70 These studies of textile workers exposed to crocidolite and amosite can thereby nolonger be used to demonstrate the risks associated with chrysotile fibres.

5.488 The seminal findings of Case et al. may cause Dr. Infante to reconsider his view – basedprincipally on the studies by Dement and by Stayner – that "chrysotile may be more potent in causinglung cancer."

(iii) Mesothelioma

5.489 On the relative risks of mesothelioma, Dr. Henderson observes that: "[T]here is generalthough not universal agreement of a differential potency between the amphiboles versus [chrysotile]for mesothelioma induction." He believes amphiboles may be greater than 60 times more likely thanchrysotile to induce mesothelioma.71 Drs. Musk and de Klerk estimate that the potency of amphibolesmay be 100 times greater. And although Dr. Infante also concedes that "amphiboles may be morepotent in causing mesothelioma", he fails to conclude from this that a distinction exists betweenchrysotile and amphibole fibres.

5.490 This distinction is also emphasized in pathology medical reference books:

"It is important to make the distinction between various forms of amphiboles and serpentines, becauseamphiboles, even though less prevalent, are more pathogenic than the serpentine chrysotile, particularlywith respect to induction of malignant pleural tumors (mesotheliomas). Indeed, some studies haveshown the link is almost invariably to amphibole exposure."72

(iv) Other Diseases

5.491 Dr. de Klerk links other asbestos-related diseases such as pleural plaques and pleuralthickening more with amphiboles than with chrysotile: "[P]leural plaques appear to be more commonamong anthophyllite workers than others while crocidolite workers have more diffuse pleural

66Henderson, see above paragraph 5.146.67Ibid.68Ibid.69Dement, J.M., Brown, D.P. and Okun, A. , Follow-Up Study of Chrysotile Asbestos Textile Workers:

Cohort Mortality and Case-Control Analyses, (1994) 26 American J. of Industrial Medicine 431.70Stayner, L., Smith, R., Bailer, J., Gilbert, S., Steenland, K., Dement, J., Brown, D., Lemen, R.,

Exposure-Response Analysis of Risk of Respiratory Disease Associated with Occupational Exposure toChrysotile Asbestos, (1997) 54 Occupational Environmental Medicine 646.

71Henderson, see above para.5.103.72See Kumar, V., Cotran, R. et Robbins, S., Basic Pathology, 6th Ed., London, Saunders Co., 1997 at

pp. 227-28.

WT/DS135/RPage 351

thickening, and benign asbestos pleurisy also seems to be more common after crocidolite exposure."Dr. Henderson also raises the issue of types of fibres in dealing with parietal pleural plaques.

Question 4(a)

5.492 Drs. de Klerk and Musk agree that the existing epidemiological data show no excess healthrisks at low-level chrysotile exposures. Dr. Henderson is not aware of exposure-response data forlow-level exposures. Dr. Infante again relies heavily on Stayner's study, a study on one single cohortof textile workers now known to be based on textile workers exposed to amphiboles as well as tochrysotile.73 Newhouse and Sullivan studied exposures to chrysotile in the manufacturing setting:"[I]t is concluded that with good environmental control, chrysotile asbestos may be used inmanufacture without excess mortality."74

5.493 Thomas et al. concluded similarly for an asbestos cement factory: "[T]hus the general resultsof this mortality survey suggest that the population of the chrysotile-cement factory studied are not atany excess risk in terms of total mortality, all cancer mortality, cancers of the lung and bronchus, orgastrointestinal cancers."75

5.494 There is clearly no increased risk of lung cancer in the friction products manufacturingindustry at levels below 356 f/ml-years. This means that there was no chrysotile-related increase inlung cancer risk for persons exposed to the equivalent of up to 8.9 f/ml for 40 years. Even if weallowed a 10-fold protection factor this would be 0.9 f/ml for 40 years for lung cancer.76 Morerecently in 1997, McDonald et al. concluded from the analysis of a cohort of 10,000 asbestos workerswith average exposures to 45 f/ml over 20 years that: "[...] from the point of view of mortality [...]exposure in this industry to less than 300 mpcf.years [approximately 45 f/ml over 20 years] has beenessentially innocuous."77 This unequivocal data comes from the longest term study of the largest groupof chrysotile workers ever conducted. A review of eight studies of cohorts exposed to chrysotile onlyled its authors to conclude: "[T]he evidence for chrysotile shows that for lung cancer andmesothelioma there exist levels of exposure below which risks are for practical purposes zero."78

Question 4(b)

5.495 According to Dr. Henderson, whether a threshold exists generally is a much-debated issue.For the case at hand, i.e. low-level exposure to chrysotile, Dr. Henderson states that: "[I]f a thresholdexists, it must lie somewhere in this area, between no exposure, low-level environmental exposure,and low-level occupational exposure." He also points out that, although no threshold has beenidentified, "[a]t the same time, no increase in risk of mesothelioma has been identified at very low-levels of exposures." Drs. Musk and de Klerk agree that the epidemiological data show an absence ofrisk at low exposure levels, but are unwilling to commit to the existence of a threshold. If there is

73See Canada's Comments to Question 3.74Newhouse, M.L. and Sullivan, K.R., A Mortality Study of Workers Manufacturing Friction Materials,

(1989) 46:3 British Journal of Industrial Medicine 176, p. 176.75Thomas, H.F., Benjamin, I.T., Elwood, P.C. and Sweetnam, P.M., Further Follow-Up Study of

Workers From an Asbestos Cement Factory, (1982) 39:3 British J. of Industrial Medicine 273, p. 275.76Berry, G. and Newhouse, M.I., Mortality of Workers Manufacturing Friction Materials Using

Asbestos, (1983) 40 British Journal of Industrial Medicine 1 at 6, p. 6.77Liddell, F.D.K., McDonald, A.D. and McDonald, J.C., The 1891-1920 Birth Cohort of Quebec

Chrysotile Miners and Millers: Development from 1904 and Mortality to 1992, (1997) 41 Annals ofOccupational Hygiene 13, p. 13.

78Browne, K. and Gibbs, G., "Chrysotile Asbestos – Thresholds of Risk" in Chiotany, K., Hosoda, Y.,Aizawa, Y., eds., Advances in the Prevention of Occupational Respiratory Diseases, Elsevier, Amsterdam, 1998at p. 306.

WT/DS135/RPage 352

agreement that low level exposures show no increased health risk, admitting the existence of athreshold is academic.

5.496 The extreme difficulty of proving a threshold scientifically is echoed by theEuropean Communities' DG XXIV Report:

"In fact, a threshold implies the demonstration that an effect does not occur at or under a given doselevel. The unequivocal demonstration (i.e. identification) of a 'negative' is tantamount to impossible."79

5.497 The corollary to the proof of a threshold is the proof of the absence of a threshold. The proofthat no threshold exists would need to explain the absence of an excess risk of lung cancer ormesothelioma in chrysotile-only cohorts, as well as the lack of any chrysotile-related increase in lungcancer mortality in workers exposed to less than 900 f/ml-years in the 10,000 miners and millersstudied in Quebec.80 Dr. Henderson does acknowledge the existence of a threshold for asbestosis inhis answer to Question 3: "Asbestosis is a lung dependent disorder with a threshold effect [...] Thereis widespread agreement that asbestosis in general is a consequence of high intensity exposure (orlower intensity but more prolonged exposure)." INSERM also supports the existence of a thresholdfor asbestosis:

"La plupart des données épidémiologiques recueillies dans des populations professionnelles exposéessuggèrent que l'asbestose cliniquement et/ou radiologiquement caractérisée n'apparaît qu'à partird'expositions suffisamment élevées [...] un seuil minimal de 25 f/ml-années a ainsi été avancé (Doll etPeto, 1985)."81

5.498 Why could there not be a threshold for other asbestos-related diseases? Dr. de Klerk assertsthat:

"[I]t is now widely believed that the risk for chrysotile workers in fibrous cement and friction productmanufacturing is so slight as to be undetectable. It is widely held that this kind of negligible risk level'threshold' exists at different levels for all types of asbestos for all relevant diseases."82

5.499 Some experts advising the EC believe there is a threshold for diseases other than asbestosis:

"It is very likely that there is a practical level of exposure below which it will be impossible to detectany excess mortality or morbidity due to asbestos. [...] Thus, it is possible that there is a level ofexposure (perhaps already achieved in the general public) where the risk is negligibly small."83

5.500 This links to Dr. de Klerk's observation that: "[T]he smaller the effect that needs to bedemonstrated, the larger the study needs to be." Dr. Infante, who dismisses the Panel's question as"moot", points out that "it is not possible to determine thresholds from epidemiological studiesbecause of the lack of statistical power to distinguish that the risk is virtually zero." Canada argues –

79DG XXIV, Opinion on a Study Commissioned by Directorate General III (Industry) of the

European Commission on "Recent Assessments of the Hazards and Risks Posed by Asbestos and SubstituteFibres, and Recent Regulation on Fibres World-Wide", Environmental Resources Management, Oxford (opinionexpressed on 9 February 1998).

80Liddell, F.D.K., McDonald, A.D. and McDonald, J.C., The 1891-1920 Birth Cohort of QuebecChrysotile Miners and Millers: Development from 1904 and Mortality to 1992,(1997) 41 Annals ofOccupational Hygiene 13.

81INSERM Report, p. 327.82de Klerk, N.H. and Armstrong, B.K., The Epidemiology of Asbestos and Mesothelioma, in Malignant

Mesothelioma, Henderson, D.W. et al., eds. Hemisphere Publishing, New York, 1992, 223 at pp. 230-31.83CEC, Report of the Working Group of Experts to the Commission of the European Communities:

Public Health Risks of Exposure to Asbestos, Oxford, Pergamon Press, 1977 cited in: WHO, EnvironmentalHealth Criteria 53 for Asbestos and Other Mineral Fibres, WHO, Geneva, 1986, p. 43.

WT/DS135/RPage 353

epidemiological data in hand – just that low-level exposures to chrysotile pose a risk that is "virtuallyzero": "un risque indétectable". Dr. Infante uses Stayner's data once again to claim that the chrysotiledata fit with a linear no-threshold model. With the new analysis on the Charleston cohort datadiscussed above, this argument does not hold.84

Question 4(c)

5.501 Drs. de Klerk and Musk agree that there is epidemiological data indicating no increased risk atlow-level exposures, but the experts believe the linear model may be appropriate. However,"[W]hether or not it is a valid method is unknown."85 According to international experts from theHealth Effects Institute-Asbestos Review (HEI-AR), such as Julian Peto, David G. Hoel andW. Nicholson, the linear model is not used for its validity, but precisely because it tends tooverestimate risk.86 Dr. de Klerk shares this view and states that the model provides a "conservativeestimate."

5.502 The limits of the linear model and the conditions under which extrapolations are made mustbe clearly set out. Extrapolations from high-level exposures and exposures to amphiboles should notbe taken at face value to ban chrysotile in today's context of low-level chrysotile-only exposures.Canada's critical view of the linear model is supported by a 1999 report by the Australian NationalIndustrial Chemicals Notifications and Assessments Scheme (NICNAS) cited by Dr. Henderson:

"There are many problems associated with low-dose risk extrapolation, such as the assumption of alinear relationship. However, as insufficient data exist to indicate threshold exposure for effect, thelinear extrapolation methodology provides a conservative worst-case scenario estimate of risk. Otherconfounding factors in estimating risks from epidemiological data are possible contamination by otherfibre types and inaccurate estimates of historical exposures."87

5.503 Not only does the linear model provide a worst-case scenario, it provides a grosslyexaggerated estimate of risk when "confounding factors", as Dr. Henderson calls them, are so clearlypresent. INSERM made extrapolations from high-level amphibole exposures to mixed fibre typeexposures, as well as from exposures in the textile industry and during the installation of low-densityproducts such as flocking.88 Amphiboles are much more potent than chrysotile, and the risks in thetextile industry cannot be compared with the risks in the high-density chrysotile products, asDr. Henderson points out in citing Boffetta: "[I]n general, the risk of lung cancer ... is highest instudies of asbestos textile workers."89

5.504 Another important consideration is the human biological defence mechanisms that arenaturally much more effective at low-levels of exposure, i.e. clearance, biopersistence and DNA repairmechanisms.90 Given these mechanisms, the reasoning behind the threshold model is both intuitively

84See Canada's comments to Question 3.85Henderson, answer to question 4(c).86Health Effects Institute-Asbestos Research, Asbestos in Public and Commercial Buildings:

A Literature Review and Synthesis of Current Knowledge (Executive Summary), Cambridge, 1991, p. 6-62.87Australia National Industrial Chemicals Notifications and Assessments Scheme (NICNAS),

Chrysotile Asbestos: Priority Existing Chemical No. 9 (Full Public Report), February 1999 at p. 72, cited byHenderson in his answer to question 4(c).

88INSERM Report, p. 213.89Henderson, paragraph 5.149 above, citing: Boffetta, P., Health Effects of Asbestos Exposure in

Humans: A Quantitative Assessment, (1998) 89 Med. Lav. 471.90Voir Holland CD, Sielken RLJ., Quantitative Cancer Modeling and Risk Assessment. Englewood

Cliffs, New Jersey: Prentice Hall, 1993; Sielken RL, Jr., Bretzlaff RS, Stevenson DE., Incorporating AdditionalBiological Phenomena into Two-Stage Cancer Models in: Spitzer HL, Slaga TJ, Greenlee WF, McClain M, eds.Receptor-Mediated Biological Processes: Implications for Evaluating Carcinogenesis. New York: Wiley-Liss,1994;237-60. Stevenson DE, Sielken Jr. RL, Bretzlaff RS., Challenges to Low-Dose Linearity in Carcinogenesis

WT/DS135/RPage 354

and scientifically sound, as well as epidemiologically validated. To illustrate this, consider thefollowing illustration: the effect of 50 fibres in the lungs will be more than five times the effect of tenfibres.

5.505 According to Sir Richard Doll, who first demonstrated the link between asbestos and lungcancer (as well as between smoking and lung cancer), "[W]e have no real ground for postulating that alinear relationship for lung cancer can be extrapolated back to the levels of dose with which we areconcerned in non-occupational settings."91 Ames and Gold are of the same view: "[l]inearextrapolation from the maximum tolerated dose in rodents to low-level exposure in humans has led togrossly exaggerated forecasts in mortality." 92 Fournier and Efthymiou are even more categorical:"[L]inear extrapolation to zero is an unscientific methodology whose social consequences are soimmense that it warrants unconditional elimination."93 INSERM acknowledges the limits of the linearmodel's application when it states that it provides nothing more than food for thought: "cetteextrapolation ne crée pas une information scientifiquement certaine, elle représente une aide à laréflexion en matière de maîtrise de risque."94

5.506 As Dr. de Klerk points out, "how one extrapolates risk assessment outside the range ofavailable data is more of a societal decision than a scientific one."

Question 4(d)

5.507 Situations where there is no increased risk at low levels of exposure have been used byStayner et al. to establish NOAELs [i.e. no observable adverse effect levels] for silica. A similarmodel is used for asbestosis. Canada believes that the use of such a model is warranted for otherasbestos-related diseases, particularly since it has been acknowledged by Dr. Musk and Dr. de Klerkthat epidemiological data exists to justify such an approach.

Question 4(e)

5.508 We concur with Dr. Henderson's view that "[t]his question iterates the issue of a thresholdexposure." Canada nonetheless notes the use by Dr. Infante of a 1992 study by Bégin et al. todemonstrate the risks related to "background levels" is erroneous. As has been pointed out by Canadain its factual arguments,95 this study is based on exposures to a mix of chrysotile and amphiboles inthe manufacturing and construction industry, and therefore is not relevant to exposures from thecurrent uses of chrysotile.

from Interactions among Mechanistic Components as Exemplified by the Concept of 'Invaders' and 'Defenders'.BELLE Newsletter 1994;3(2):1-8. Stevenson DE., Dose-Response Studies of Genotoxic Rodent Carcinogens:Thresholds, Hockey Sticks, Hormesis or Straight Lines? - Comment on the Kitchin and Brown paper, BELLENewsletter 1995;3(3):14-15.

91Doll, R., Mineral Fibres in the Non-Occupational Environment: Concluding Remarks, in Bignon, J.,Peto, J. and Saracci R., eds., Non-Occupational Exposure to Mineral Fibres, IARC Scientific Publication No 90,1989, pp. 516-17.

92Ames, B.N. et Swirsky Gold, L., Causes and Prevention of Cancer: Gaining Perspectives on theManagement of Risk, in Risks, Costs, and Lives Saved: Getting Better Results From Regulation?, New York,OUP, 1996, p.6.

93Fournier, E. and Efthymiou, M.-L., Problems with Very Low Dose Risk Evaluation: The Case ofAsbestos, in What Risk?, p.49.

94INSERM Report, p. 239 and 414.95 See above Section III.A.5.

WT/DS135/RPage 355

Question 5(a)

5.509 Clearly the answers given by the four experts are based on their concept of what is meant bycontrolled use. It is also evident that the controlled use concept as espoused by Canada was not theapproach that resulted in their answers. We must therefore respectfully disagree with the answersgiven by the experts in respect of controlled use of chrysotile and high-density chrysotile containingproducts. The fact that they agreed that controlled use of chrysotile and high-density chrysotileproducts is feasible at some points of the life cycle, but not in others, suggests that they are not farfrom the view of Canada. The only difference is that Canada believes that the experts misunderstandthe controlled use principle and that, as properly understood and implemented, use can be controlledthroughout the full life cycle of high-density chrysotile containing products. The basis for our view,with supporting evidence, is set out below.96

(i) Canada's understanding of the "Controlled use" principle

5.510 The Canadian government's review of the experts' reports and answers to the questions posedby the Panel reveals that there is one crucial issue, which seems to override all other issues. This isthe question of whether the application of the controlled use principle is feasible and credible in allstages in the life cycle of a product. While there is a reasonably high degree of agreement amongexperts that controlled use can be a reality in the mining and manufacturing sectors, serious doubts areexpressed that controlled use can be applied in a few sectors of use − installation, maintenance anddemolition. However, the basis for this view is not documented, except by Dr. Infante andDr. Henderson.

5.511 By "controlled use", the Canadian government means "stewardship" based on the total lifecycle. This is outlined in the document The Mineral and Metals Policy of the Government of Canada:Partnerships for Sustainable Development.97 With regard to asbestos, this "controlled use" is basedon the following general principles:

• Only the chrysotile variety is used;

• only a limited number of well-defined product applications, where it has beendemonstrated that they can be handled safely, are allowed (i.e. where the fibres areencapsulated in a matrix such as cement, bitumen, plastic, resin, etc.);98 and

• new product applications may be introduced only after a strict evaluation to ensurethat a certain level of fibre release is not exceeded during its life cycle.

96Canada notes that the "controlled-use" approach has been endorsed by the WHO in its 1998

Environmental Health Criteria 203: Chrysotile Asbestos, p. 144. "Control measures, including engineeringcontrols and work practices, should be used in circumstances where occupational exposure to chrysotile canoccur. Data from industries where control technologies have been applied have demonstrated the feasibility ofcontrolling exposure to levels generally below 0.5 fibres/ml. Personal protective equipment can further reduceindividual exposure where engineering controls and work practices prove insufficient."

97NRCAN, The Minerals and Metals Policy of the Government of Canada: Partnership for SustainableDevelopment, Public Works Canada, 1996. Canada notes that the "controlled-use" approach to regulatingchrysotile asbestos is well researched as evidenced in the studies and conclusions referred to by Canada in itsfactual arguments (see above Section III.A.6).

98To illustrate this point, examples of "controlled use" of friction products and asbestos-cement aredetailed in Appendices A and B respectively to these comments. (These Appendices can be found in Annex IVto this Report).

WT/DS135/RPage 356

5.512 With regard to the downstream use sectors, "controlled use" implies that alldistributors/manufacturers of asbestos will be required to have an import permit. This permit will bewithdrawn if the company does not meet the following commitments:

• To distribute its products only to companies (users) licensed to purchase theseproducts. Those companies must have workers trained and licensed to installproducts, and must be in compliance with regulations. Approved users shall not resellto third parties, and any unused materials must be returned to the manufacturer;

• to provide a list of users of products to the responsible government agency;

• to provide products cut to specification and to establish centres equipped to cut theproducts to size, and where persons cutting the products are trained and are licensedto work with asbestos; and

• to police the downstream users in co-operation with the government. The productmanufacturer visits, monitors and reports on the performance of the downstream usersat regular intervals. There are penalties for failing to provide this productstewardship.

5.513 While high-density products in most countries are not considered to pose any occupational orenvironmental health risk, disposal should only be undertaken by approved and appropriately trainedpersons.

5.514 Dr. Infante's description of the permissible exposure limit for chrysotile asbestos, as well asprogrammes or standards that recommend or require specific engineering control, work practices,training and education and personal protective equipment to control exposures to asbestoscorresponds, to some extent, to Canada's approach. Dr. Infante seems to suggest that because someworkers do not comply with standards and regulations on controlled use in the United States,controlled use is not feasible. As explained in Appendix A on friction material and Appendix B onasbestos cement, the controlled use approach can minimize, if not eliminate, workers' non-compliance.99

5.515 Canada does not propose that any chrysotile products produced, sold or used without theimplementation and enforcement of very stringent control procedures. Taking into account the typesof products being manufactured and used in France at the time of the ban, Canada does not advocatere-introducing any product that cannot be handled according to the safety criteria outlined above.Canada is not advocating the introduction anywhere in the world of manufacturing facilities ofproducts for which the technology does not exist to protect workers from exposure to chrysotile atlevels where risks would be above epidemiologically based practical thresholds.

5.516 The experts have indicated that the level of exposure is such that they are not concerned aboutasbestos-related disease for persons living in buildings containing chrysotile asbestos products,including friable insulation. As none of the chrysotile products that will be used in the future arefriable, this conclusion would be further reinforced. If the procedures envisaged under the "controlleduse" policy are followed by licensed practitioners, the public will not be placed under any practicallydeterminable increased risk of disease as a result of the manufacture and use of chrysotile containingproducts. Unlike friable insulation products where janitorial staff, electricians, carpenters, and othersmay be required to work regularly in an environment where exposures to asbestos would occur, thenature of the high-density products will ensure that exposures are a much rarer event.

99Appendices A and B can be found in Annex IV to this Report.

WT/DS135/RPage 357

5.517 Canada recognizes that the clock cannot be turned back. Friable mixed products produced inthe past are now in place, and trades such as electricians or telephone engineers face situations wherethe potential health risk from exposure is considerably greater than any additional risk that new high-density chrysotile products would present. It is evident that the protection of workers who come intocontact with friable products must be assured by the responsible jurisdictions through training in tradeschools, appropriate information programmes by unions, and by governments and employers ensuringthat the appropriate equipment and tools are made available to workers.100

5.518 Regarding high-density products, Canada believes that no less stringent measures should berequired, even though the evidence shows that the risk from exposure to high-density chrysotileproducts is minuscule compared to the risk from friable products, in many cases containing mixturesof chrysotile and amphibole fibres. Furthermore, in the absence of sound scientific data to thecontrary, the same criteria should be applied to the handling of all products in which respirable fibres,including asbestos substitutes, may be released.

(ii) International Standards

5.519 None of the experts acknowledges that controlled-use of chrysotile asbestos cement productsand other high-density chrysotile products stems from international standards. Dr. Infante even deniesthe existence of international standards on controlled-use of high-density chrysotile products. Canadawishes to remind the Panel that international standards, as the term is defined in the Agreement onTechnical Barriers to Trade, do exist. Regulatory developments on asbestos fibres have been guidedby ILO Convention 162 concerning Safety in the Use of Asbestos.101 ILO convention 162 providesfor: (i) the prescription of adequate engineering controls and work practices; (ii) the prescription ofspecial rules and procedures for the use of asbestos or certain types of asbestos or products containingasbestos or for certain work processes; (iii) where necessary to protect the health of workers andtechnically practicable, the replacement of asbestos or of certain types of asbestos by other materialsor the use of alternative technology scientifically evaluated by the competent authorities as harmlessor less harmful; and (iv) total or partial prohibition of the use of asbestos or of certain types ofasbestos in certain work processes.102

5.520 The Code of Practice on Safety in the Use of Asbestos of the International Labour Officereferred to by Canada in all its submissions is another international standard on controlled-use.103 Theobjects of the Code are: (i) to prevent the risk of exposure to asbestos dust at work; (ii) to preventharmful effects on the health of workers arising from exposure to asbestos dust; and (iii) to providereasonably practicable control procedures and practices for minimising occupational exposure toasbestos dust. To do so, the Code gives detailed guidance on the limitation of exposure in respect ofasbestos cement and friction materials. Finally, Canada has referred the Panel to InternationalStandard ISO 7337: Asbestos Reinforced Cement Products − Guidelines for On-Site WorkPractices.104 This international standard gives guidelines for tools and working methods to be used onsite with a view to maintaining the dust emission at the lowest practicable level. It applies to asbestos-cement products.

100See Section III.A.5 of this Report and Camus M., L'amiante et les risques pour la santé, April 1999.101Conférence internationale du travail, Convention concernant la sécurité dans l'utilisation de l'amiante

(Convention 162), adoptée le 24 juin 1986, and Recommandation concernant la sécurité dans l'utilisation del'amiante (Recommandation 172), adoptée le 24 juin 1986.

102According to Canada, the emphasis of ILO Convention 162 is on controlled-use and not on productprohibitions. The Convention calls for two specific prohibitions: crocidolite and all products containingcrocidolite, and sprayed-on applications of asbestos.

103Recueil de directives pratiques du BIT sur la sécurité dans l'utilisation de l'amiante, Organisationinternationale du travail, Genève, 1984.

104ISO, standard ISO-7337 1984.

WT/DS135/RPage 358

5.521 The ILO Convention 162 and the Code of Practice on Safety in the use of chrysotile should besupplemented by a national policy on responsible use based on the recognition and acceptance of theprinciples that both international standards set forth.105 As explained above, the objective ofresponsible use is to limit the handling of chrysotile to companies that comply with the nationalregulations or that have submitted action plans and formal commitments in writing with a view tobringing their activities into line with these regulations.

Question 5(b)

5.522 The experts recognize that training could be achieved in the manufacturing sector, where thereis a small and cohesive workforce, but assert without support that it cannot be achieved in theconstruction sector, where there is a large and non-cohesive workforce. Dr. Infante wrongly equatesnon-compliance with regulated training requirements to non-feasibility of training for controlled-useof chrysotile asbestos.106

5.523 In Europe, as in other countries, there are now requirements for training workers. In Canada,both levels of government require training at all workplaces. It is possible for training to be madeavailable by industry. In fact, information and training is one of the most important elements of acompany's preventive control programme. In line with the controls suggested at paragraphs 5.511and 5.512, France could require through legislation that all construction workers handling asbestosproducts attend training sessions. France could also require that only designated, properly trainedworkers be allowed to work with those asbestos products that need to fall under a controlled regime.

5.524 During manufacture, controls such as wet processes and exhaust ventilation, essentiallyeliminate all exposure. On the work site, process changes are reduced by the industry manufacturingproducts requiring no, or virtually no, modifications on site. The controlled use approach includes theuse of pre-cut and pre-drilled asbestos cement products, and provides for designated locations wherechrysotile asbestos cement sheets or pipes are cut and drilled and where the appropriate controls are inplace. The monitoring process is similar to that for other workplaces: all complaints are submitted togovernmental inspectors for evaluation. The supplier has the responsibility for ensuring that allcompanies to which they supply have in place the proper equipment and training to ensure safe use ofthe product throughout its life cycle. Finally, the removal of high-density chrysotile products iscarried out in accordance with government codes.

Question 5(c)

5.525 Both Dr. Henderson and Dr. Infante agree that, in many situations, when standards areproperly applied, it is possible to maintain exposure below 0.1 f/ml. Also, as explained inAppendix A on the friction industry and Appendix B107 on the asbestos cement industry, experienceshows that a level below 0.1 f/ml can be achieved because the technology and work practices exist tocontrol exposure during manufacture. No guarantee can be offered that there would never be asituation in which 0.1 f/ml might be exceeded as a peak exposure. However, there is no evidence thatoccasional peak exposures increase the risk of lung cancer or mesothelioma in chrysotile exposedworkers. For example, the health experience of brake mechanics, i.e. no evidence of an increased riskof mesothelioma or lung cancer, is based on exposures that involved peak exposures, such as occurredduring the blowing out of brake wear debris and the occasional grinding of brake linings. Theseoperations involved short exposures above 0.1 f/ml. The actual concentrations associated with various

105Memorandum of Understanding between the Government of Canada and the Asbestos Industry on

Responsible-Use of Chrysotile Asbestos, 1997.106Canada notes that it should be recalled that his basis for risk assessment is based on the textile

industry.107See Annex IV to this Report.

WT/DS135/RPage 359

tasks have been reported by Kauppinen and Korhonen,108 and by Rödelsperger.109 In spite of theseshort-term peak exposures, the average exposure of auto mechanics was less than 0.05 f/ml.

5.526 A person repairing their own brakes periodically nowadays (using disc brake pads mainly)would have extremely low cumulative exposures compared to full time auto mechanics and there is noreason for them to have any, even short term, exposures exceeding 0.1 f/ml. The risks associated withcumulative exposure to chrysotile at these levels would not be epidemiologically detectable forhandymen handling friction or asbestos cement products.

5.527 Rödelsperger110 made dust measurements on about 40 buildings sites in Germany. Hereported peak exposures of more than 100 f/ml in the vicinity of a grinding machine used to cutasbestos cement sheets. However, when he used the standardized work histories of 61 roofers, whohad a mean duration of exposure of 16 years, he found that their mean cumulative exposurewas 1.6 fibre-years/ml. These measurements were made 20 or more years ago, with the products andtechnology available then and for regular construction workers. It is evident that even under thesecircumstances, lifetime cumulative exposures were low. Thus, a handyman, even if he did not takeproper precautions would still have a low cumulative fibre exposure because peak exposures are ofshort duration and he would be at a very low, undetectable risk of health effects.

5.528 It is generally agreed that at the levels of exposure associated with the use of the modern high-density products, they would not even put a full-time worker at increased risk of asbestosis and,therefore, this would not be of concern for a handyman working occasionally with the product. It hasbeen amply demonstrated that the risk of lung cancer increases with increasing cumulative lifetimeexposure that combines duration and level of exposure. A person exposed at 0.1 f/ml for 40 years hasa cumulative lifetime exposure of 4 f/ml-years. If that person worked on a project only once eachweek for four hours for 40 years, he would not achieve the same lifetime exposure unless he wasexposed to 1 f/ml continuously for the four hours of exposure every time he was exposed for 40 years.Thus occasional peak exposures of a few minutes contribute very little to cumulative lifetimeexposure which is important in evaluating the risk of chronic diseases such as lung cancer ormesothelioma.

5.529 Gardner111 found no increased risk of lung cancer or other asbestos-related disease in achrysotile asbestos cement plant where exposures were less than 1 f/ml. This was in a cohort ofworkers employed between 1941 and 1983. It is evident that any risk would have been well below thedetection limit at 0.1 f/ml. A study of chrysotile cement production workers by Thomas112 andNeuberger & Kundi113 identified no chrysotile-related increased risk of lung cancer and Weill,114 whilereporting an increased risk of lung cancer in asbestos cement workers, found the increased risk only in

108Kauppinen, T. and Korhonen, K., Exposure to Asbestos During Brake Maintenance Of Automotive

Vehicles by Different Methods, (1987) 48 Am. Industr. Hyg. Assoc. J, pp. 499-504.109Rödelsperger, K. et al., Asbestos Dust Exposure During Brake Repair, (1986) 10 American Journal

of Industrial Medicine, pp. 63-72.110Rödelsperger, K., Woitowitz, H.J. and Krieger, H.G., Estimation of Exposure to Asbestos-Cement

Dust on Building Sites, in Biological Effects of Mineral Fibres, Vol. 2, J.C. Wagner Editor, 1980, InternationalAgency for Research on Cancer: Lyon, pp. 845-853.

111Gardner, M.J., Winter, P.D., Pannett, B. and Powell, C.A., Follow-Up Study of WorkersManufacturing Chrysotile Asbestos Cement Products, (1986) 43 British J. of Industrial Medicine, pp. 726-732.

112Thomas, H.F., Benjamin, I.T., Elwood, P.C. and Sweetman, P.M., Further Follow-Up Study ofWorkers from an Asbestos-Cement Factory, (1982) 39 British Journal of Industrial Medicine, pp. 273-276.

113Neuberger, M. and Kundi, M., Individual Asbestos Exposure: Smoking and Mortality − A CohortStudy in the Asbestos-Cement Industry, (1990) 47 British Journal of Industrial Medicine, pp. 615-620.

114Weill, H., Biological Effects: Asbestos-Cement Manufacturing, (1994) 41 Ann. Occup. Hyg.,pp. 533-538.

WT/DS135/RPage 360

those with asbestosis. In this study, there was little evidence of asbestosis below 30-40 f/ml-years ofexposure. This is about 0,75-1 f/ml continuous exposure for 40 years. Thus, there is little evidence tosupport a detectable increase in risk of lung cancer in workers with a 40 years cumulative lifetimeexposure at 4 f/ml-years.

5.530 Any risk estimates obtained by linear extrapolation from high exposures to such lowexposures are somewhat hypothetical and both Lash115 and Camus116 have shown that the riskestimates made by the U.S. Government have overestimated lung cancer risks.

Question 5(d)

5.531 Canada disagrees with the views of Dr. Henderson117 and Dr. Infante that controlled use ofchrysotile asbestos is not feasible for workers involved in the construction trade and that service andmaintenance workers such as carpenters, plumbers, and electricians will experience peaks ofexposures to asbestos that place them at risk. The nature of high-density chrysotile asbestos productsis such that few of the trades listed above will ever need to work on the products, with the possibleexception of demolition workers. Again, there is evidence that during demolition exposureconcentrations associated with chrysotile asbestos cement products is very low.118 Today, withchrysotile cement products and controlled use procedures, health risks become insignificant.

5.532 Recommended installation methods can eliminate the need to cut or drill into chrysotile-basedproducts at construction sites, since those products are distributed in a variety of pre-cut and pre-drilled sizes, according to buyers' specifications. In fact, many asbestos cement products are pre-formed ready for use. They are factory-made to the correct size and shape including holes so that aminimum of on-site preparations is needed. Once installed, chrysotile asbestos cement pipes arebelow ground and pose no risk to workers. Even if dug up, they pose no risks unless comminuted,ground or sawed, and, when this is necessary, the use of appropriate tools and controls will keep therelease of dust and exposure well within the level considered safe by the WHO. Chrysotile asbestoscement sheets are used for roofing and exterior building walls. Once installed, there is no need tomodify the roof until the life of the product is over. Similarly, there is no need to modify chrysotileasbestos sheets used as walls once they have been installed. The product can be painted without fibrerelease.

5.533 Chrysotile cement products are unlikely to release fibres into the environment or breathingzones of workers such as janitors, plumbers, electricians, repair men, etc., unless these workers haveto actually cut or drill the product. Unlike insulation products, there will rarely be a need for anyoneto perforate, saw, or grind installed chrysotile cement products. Where cutting or drilling is required,hand tools and low speed power tools are recommended in combination with wetting to keep dustlevels to a minimum. Dust levels for various types of on-site working have been measured both inlaboratories and in the field and these facts showed that risks could be maintained below detectionlimit.

115Lash, T.L., Crouch, E.A.C. and Green, L.C., A Meta-Analysis of the Relation between Cumulative

Exposure to Asbestos and Relative Risk of Lung Cancer, (1997) 54 Occupational and Environmental Medicine,pp. 254-263.

116Camus, M., Siemiatycki, J. and Meek, B., Nonoccupational Exposure to Chrysotile Asbestos and theRisk of Lung Cancer, (1998) 338 N. Eng. J. Med., 1565.

117Henderson, answers to Questions 5(a) to (d).118Hoskins J.A., Chrysotile in the 21st Century, UK, 1999, p.12.

WT/DS135/RPage 361

Question 5(e)

5.534 Dr. de Klerk and Dr. Musk wrote that efficiency of controlled-use in the case of homehandymen is outside their area of expertise. However, both Dr. Henderson and Dr. Infante haveconcluded that it is not possible to control exposure to chrysotile asbestos high-density products innon-occupational circumstances (occasional interventions by home handymen). Neither bases hisconclusion on data. Dr. Henderson adds to his answer that although such risks are not quantifiablebecause of absence of data, these risks must be very small for lung cancer and mesothelioma, and non-existent for asbestosis.

5.535 Controlled use will reduce and even eliminate risks. The risk of chrysotile-related healtheffects is tied to cumulative exposure, that is, duration and level of exposure. Rarely will anindividual under non-occupational circumstances achieve the exposure of a full-time worker.Occasional uncontrolled exposures for a handyman would not result in appreciable cumulativeexposure. Data published by Brown119 showed time-weighted average (TWA) levels duringdemolition of weathered asbestos-cement roofing between 0.3 and 0.6 f/ml. One can likely guess thata handyman would not practice such an activity more than 40 hours in 25 years. This would averageout to a TWA of 0.015 f/ml for the year of this activity, and a TWA of 0.0006 f/ml each year of theworker's adult life. This is 1 million times less than past asbestos workers are. It is equivalent toexposure levels in schools containing ACM.120

5.536 Based on INSERM121 and HEI-AR risk tables, which are based on mixed asbestos exposures,the resulting lifetime cancer risk would be between 10 and 20 in a million depending on the timeoccurrence of this exposure scenario. More accurately however, the lifetime risk would be near zeroper million, based on chrysotile friction workers who were exposed to similar fibres (species anddimension wise), and about 1 in a million, based on the risks of past chrysotile miners and millers.The casual user of a high density product, even if the product were weathered, is not likely to be atany increased risk of an asbestos related disease. If the supplier follows through on the requirementsof controlled use, the casual purchase of chrysotile asbestos-cement products by the handyman willnot be possible. However, there is probably no way of stopping any individual from doing somethingto any product if they can obtain it. This is a problem that exists for any product many of which poseserious health risks if abused.

Question 6(a)

5.537 Canada respectfully disputes the conclusions of the experts regarding the risk from substitutefibres, and with respect to one expert, the ability of substitutes to serve as suitable replacements tochrysotile. Canada notes that the treatment of the issue by two of the experts is terse, comprising onlyseveral sentences. To their credit, Drs. de Klerk and Musk indicate that the use and control ofsubstitute fibres is not within their areas of special expertise. They offer some responses nonetheless.Canada is concerned, in particular, by their lack of familiarity with the relevant studies and actualmodes of production, use and disposal of substitute fibres. For example, they apparently are unawareof research conclusively demonstrating the significant health risks from exposure to refractoryceramic fibres (RCF), which are discussed below.

5.538 This concern applies to Dr. Infante as well. Dr. Infante further appears unaware of (orignores) recent research demonstrating that chrysotile is less biopersistent than many substitute fibres.

119Brown, S.K., Asbestos Exposure During Renovation and Demolition of Asbestos-Cement Clad

Buildings, (1987) 48 Amer. Ind. Hyg. J., pp. 478-486.120Health Effects Institute − Asbestos Research, Asbestos in Public and Commercial Buildings: A

Literature Review and Synthesis of Current Knowledge (Executive Summary), Cambridge, 1991, pp. 1-11.121INSERM Report.

WT/DS135/RPage 362

Dr. Infante also ignores the population the experts agree are most at risk from exposure to any fibre –tradesmen – when he concludes (without support or, even, explanation, Canada notes) that the "natureof the production process makes substitutes more amenable to control than asbestos fibres."Assuming he is not again conflating chrysotile with amphiboles when referring to "asbestos," hispoint, were it true, would be irrelevant. The experts agree that chrysotile and chrysotile products canbe safely mined and produced. The key is exposure to tradesmen. And, for that exposure, norationale exists suggesting that the ability to impose effective controls differs based on the type of thefibre.

5.539 Dr. Henderson, for his part, recognises that, as with all fibres, the pathogenicity of substitutesis defined by the "3Ds" (dimension, dose, durability). He seems also to understand that, due to the(lack of) historical use of substitutes, we cannot fully know the risks of using them.122 However, hethen seems to ignore the importance of these facts.

5.540 All of the experts fail to take into account several very important factors. First, the chrysotileproducts at issue in this proceeding are quite few. Second, the exposure levels during themanufacture, use and disposal of these products are extremely low. Third, the data demonstrate thatthese few products have been and can be used without detectable health effects in humans. Moreover,in order to assess whether a substitute is safer to use than chrysotile in a product: (i) it is fundamentalthat the characteristics of the fibres being compared be those of the fibres as they are used in theproduct or as they are released from the product throughout the product's lifecycle; (ii) it is essentialthat data on at least the key parameters (exposure, biopersistence and dimensions) be available tomake this assessment. Unfortunately, the experts have not addressed these topics. In short, theexperts have based their opinions on very limited, if any, data. While the experts reach conclusionsthat various substitutes (PVA fibres, glass fibres, cellulose and para-aramid fibres) are safer to usethan chrysotile, they provide no systematic comparison of risks and very limited, questionablescientific data in support of their opinions.

5.541 Canada presents below a survey of the studies and concepts that the experts ignored. Thesestudies give a picture of risks from substitute fibres starkly different from that suggested by the Panel'sexperts. As demonstrated below, the situation concerning risk from substitute materials is as Canadaset out in its factual arguments.123

(i) The Fibres to Compare

5.542 Experimental data for a wide range of fibres have shown that the physical characteristics(diameters, lengths, density) of fibres are important in determining their respirability, when they aredeposited in the respiratory system, and their capacity to induce fibrosis and cancers. Further, the riskof effects also depends on dose (exposure). Thus, differences in the risk of disease in variousindustrial sectors would be expected to occur because of differences in these, as well as other factors.As the characteristics of chrysotile and any substitute fibres are likely to be dictated by the product inwhich they are used, it is not appropriate to assess the risks associated with friction products orasbestos cement products using data from other industrial sectors. The data that should be availableand used for the purpose of comparing risks should be those for the fibres as used in the specificproducts under review. Canada's presentation proceeds on this basis.

5.543 Davis124 pointed out that while materials like wool, cellulose and other fibres have in somecases been used for many years, they are now being used in quite different applications, about which

122Henderson, response to Question 6(b).123See above Section III.A.6.124Davis, J.M.G., The Toxicity of Wool and Cellulose, (1996) 12 J. Occ. Health and Safety Australia and

New Zealand, pp. 341-344.

WT/DS135/RPage 363

knowledge is very limited. As a consequence, the characteristics of the fibres used in the newerapplications may not be the same as those in the conventional products manufactured in the past.Such changes can modify the respirability and biological activity of the materials. There is a furthercomplication for substitutes that is not addressed by the experts. This is the fact that substitution doesnot always involve replacement of chrysotile by a single fibre, but often by several different materialsor substitute fibres. For example, cocktails of fibres are needed to meet technical requirements infriction products. In addition, when substituting for chrysotile, other materials such as silica or otherfibres, fire retardants or biocides must often be added. These agents may themselves be toxic orcarcinogenic, and may act synergistically.

(ii) The Outcomes to be Measured

5.544 While it is reasonable to compare the risks of lung cancer and mesothelioma between thevarious fibre types, it must be remembered that different sized fibres may lead to fibre deposition atdifferent locations in the respiratory system. For example, if more fibres of one material than anotherare likely to be deposited in the nasal passages, one should consider the possibility of an increasedfrequency of nasal cancer in evaluating the substitute. Dr. Infante mentioned the increased risk ofnasal cancer in woodworkers, which has been well established.125 This might raise a questionconcerning the sources of the cellulose used as a substitute and whether controls are in place to avoidexposure to cellulose from woods that have caused such cancers. Also, some materials may causedangerous allergic responses. Certain glass fibres cause skin irritation. Harrison126 notes that thereare indications of an accumulation of oligomers in the kidney in some circumstances, so that attentionshould be given to the molecular weight of PVA used "especially if a smaller diameter material wereto be produced."

5.545 In considering risks, the composition of the dusts and fibres to which workers are exposedwhen handling the "raw substitute" materials, manufacturing the product, cutting, grinding,manipulating or disposing of the product also must be considered. For example, it is important toknow whether the fibres of para-aramid, PVA or cellulose are opened (fibrillised) or comminutedduring preparation or manufacture of the product? Does manipulation, sawing or drilling of theproduct give rise to narrower diameter respirable "fibres" such as result with polyester fibres duringweaving? Do these fibre fragments have biological significance? What are actual use concentrations?It must be remembered throughout that there exists a substantial body of information concerningchrysotile. Unfortunately, in the case of substitutes, there are rarely any human epidemiological dataavailable and even experimental data are limited. Perhaps this fact led Dr. de Klerk to conclude that"[G]iven the comparative lack of knowledge about the health effects of substitute materials, thecontinued use of chrysotile under [controlled] circumstances seems sensible.127

(iii) The Essential Data

5.546 The data that need to be compared in an evaluation of the relative safety of chrysotile andsubstitutes include the following:

• Epidemiological data that provide direct evidence of the risks associated with theproducts.

125Infante, answer to Question 6(c).126Harrison, T.W., Levy, W.S., Patrick, G., Pigott, G.H. and Smith, L.L., Comparative Hazards of

Chrysotile Asbestos and its Substitutes: A European perspective, (1999) Environmental Health Perspective, 107.127de Klerk, N.H. and Armstrong, B.K., The Epidemiology of Asbestos and Mesothelioma", in

Malignant Mesothelioma, Henderson, D.W. et al., eds Hemisphere Publishing, New York, 1982, p. 231.

WT/DS135/RPage 364

• Experimental data by inoculation of fibres or by inhalation experiments inexperimental animals.

• Dimensions of fibres in the respirable airborne dust during the manufacture of theproduct.

• Dimensions of fibres in the respirable airborne dust during the use of productscontaining the fibre.

• Dimensions of fibres in the lungs of workers engaged in the manufacture of productscontaining the fibre.

• Dimensions of fibres in the lungs of persons exposed during the use of productscontaining the fibre.

• Dimensions of fibres in the respirable airborne dust and in the lungs followingexposure during the disposal of the fibre or products.

• The biopersistence of the fibres in humans and animals.

• The cumulative exposure (i.e.: concentration x time) of workers engaged in all phasesof manufacture, use and disposal of the product.

• Data on alterations or modification of the fibres chemically, physically andbiologically during their life cycle that might affect their potential to cause healtheffects.

5.547 Even if one were to narrow the requirements to a smaller number of key parameters such asfibre dimensions, biopersistence and exposure-response, the available data are still inadequate toprovide a credible basis for an adequate comparison. Thus, the unqualified wholesale affirmation that"substitutes are safer than chrysotile" is not well founded and potentially very dangerous. Forexample, prior to the finding of a very high risk of mesothelioma for persons exposed to very lowconcentrations of fibrous zeolite erionite in Turkey, there had been no indications world-wide thatsuch fibres might after 30 years from first exposure at such very low levels produce such a high rate ofmesotheliomas in humans. In South Africa, crocidolite had been used for about 60 years beforeWagner128 reported that mesotheliomas were associated with crocidolite exposure. In humans,mesotheliomas do not occur until 40-60 years after first exposure. Thus, caution is needed in theabsence of data regarding substitute fibres. As one expert stated: "better the devil you know" than thedevil you do not know.129

Question 6(b)

(i) Dimensions

5.548 The experts present no data that show that the dimensions of all fibrous substitutes are outsidethe respirable size range during the substitute product's lifecycle. This is because no such data exist.

128Wagner, J.C., Newhouse, M.L., Corrin, B., Possister, C.E. and Griffiths, D.M., Correlation between

Fibre Content of the Lung and Disease in East London Asbestos Factory Workers, (1988) 45 British J. ofindustrial Medicine, 305.

129de Klerk, answer to Question 6(a).

WT/DS135/RPage 365

(ii) PVA & Aramid Fibres

5.549 Views from the experts appear to be mixed. Dr. Henderson quotes Harrison's review statingthat PVA and aramid fibres are too large to be respirable. Dr. de Klerk states that all substitutesexcept glass (cellulose, aramid, PVA) produce a larger proportion of non-respirable fibres thanchrysotile but respirable fibres are similar for all substances. Dr. Musk offers no opinion. Dr. Infantestates that PVA fibres are "mostly" in the size range 10-16 µm and aramid fibres 10-12 µm. However,he notes, quite correctly, that, as mentioned below, the aramid fibres can and do split into fibrils ofabout 0.2 µm in diameter.

5.550 In assessing fibre respirability, none of the experts accounts for the fact that respirabilitydepends on density as well as fibre diameter. The densities of PVA and para-aramid fibres are bothconsiderably less that that of chrysotile. This means that much larger diameter substitute fibres wouldbe respirable. In fact, the upper limits of diameters that are respirable for these fibres, as reported byHarrison130, are approximately 7 µm and 6-7 µm respectively. The equivalent upper level diameter forchrysotile is about 3-3.5 µm. Thus, fibres of much greater diameter can penetrate into the alveolarregion of the lung. A review of the available information in the literature is that there is a generalopinion without data that the respirable fraction of PVA fibres is small. However, there do not appearto be any data on the dimensions of airborne fibres during mixing with cement or other materials, or asreleased from the products during processing and use.

(iii) Glass and Cellulose Fibres

5.551 As far as cellulose and glass fibres are concerned, none of the experts provided any actualmeasurement data on the sizes or respirability of the fibres. Also, the dimensions of fibres at variousstages of the processing, use and disposal of cellulose have not been reported. The actual dimensionsof fibres in the airborne dust will depend on the specific glass fibres used and how they were prepared.

(iv) Biopersistence

5.552 It is well known that biopersistence is a key parameter. Indeed, the human evidence forchrysotile indicates that it is likely to be one of the main reasons why chrysotile is less dangerous thanthe amphiboles in respect to mesothelioma risk. This is clearly recognized by three of the fourexperts, as well as by INSERM.131

(v) Cellulose

5.553 Drs. Infante, Henderson and de Klerk recognise that cellulose is durable in the lung. In fact,the data show that some cellulose fibres have half lives of about 1000 days in the lung, which aremany times longer than even those published data for amphibole fibres, much less chrysotile fibres.132

(vi) PVA

5.554 Dr. Musk and Dr. Henderson had no comment on PVA durability. Dr. de Klerk presented nodata, but expressed the view that PVA was less durable than chrysotile. Davis, in a review in 1998,found no published data on the biopersistence of PVA fibres. Data was not published until 1999,

130Harrison, T.W., Levy, W.S., Patrick, G., Pigott, G.H. and Smith, L.L., Comparative Hazards of

Chrysotile Asbestos and its Substitutes: A European perspective, (1999) Environmental Health Perspective, 107.131See Canada's comments on Question 3.132Muhle, H., Ernst, H. and Bellman, B., Investigation of the Durability of Cellulose Fibres in the Rat

Lungs, (1997) 41 Ann. Occup. Hyg., pp. 184-188.

WT/DS135/RPage 366

when Harrison [1999] reported that PVA "will degrade very slowly, if at all in the lung."133 There donot appear to have been any systematic studies of the biopersistence of PVA fibres, a crucialparameter in assessing the hazard associated with PVA fibres.

(vii) Para-aramid Fibres

5.555 Based on a study by Searl,134 who compared chrysotile and para-aramid fibres, the generalview of the experts is that para-aramid fibres are less biopersistent. However, Searl failed to check thelung tissue to confirm that the retained fibres were chrysotile. Based on studies using a standardprotocol, Dr. David Bernstein has found that the biopersistence of chrysotile is in fact less than that ofpara-aramid fibres.135

(viii) Glass Fibres

5.556 Drs. Musk, de Klerk and Henderson presented no data on the biopersistence of glass fibres.Dr. Infante, without identifying the specific glass fibres, reports that glass fibres are less biopersistentthan chrysotile. In fact, the recent work by Dr. Bernstein, in which the same protocol was used as forsynthetic fibres, found that long [i.e.: > 20um] pure chrysotile fibres are removed from the lung fasterthan most, if not all, of the glass fibres reported in the published literature.136

(ix) Chrysotile

5.557 As far as chrysotile is concerned, it is well accepted that chrysotile is readily removed fromthe lung. This is why the lungs of chrysotile millers and miners, exposed to chrysotile, have beenfound at autopsy to contain more tremolite (an amphibole asbestos mineral) than chrysotile.137 Thechrysotile cleared, but the tremolite fibres remained in the lung because of their much greaterbiopersistence. There are various estimates of the half time for chrysotile clearance. Oberdöerster138

studied baboons and estimated a 90-110 day half time for chrysotile fibres. The study by Searl wasmentioned above. The estimates by Dr. Bernstein are even shorter (< 10 days).139 For directcomparison purposes, the clearance rates for fibres in the same ranges of dimensions must be studied.In addition, it is crucial that the fibres be tested using the same methodology. The studies byBernstein best fit these criteria and show that, size for size, chrysotile has a very short half-life.

Question 6(c)

(i) Exposure-Response

5.558 In the absence of exposure-response data, it is not possible to quantify the risks associatedwith the various fibres. The question to be addressed is not: is one material more dusty than another?Nor is it: is the concentration higher when working with one material compared to another? Rather,

133Davis, J.M.G., The Biological Effects of Fibres Proposed as Substitutes for Chrysotile Asbestos:

Current State of Knowledge, 1998.134Searl, A., Clearance of Respirable Para-Aramid from Rat Lungs: Possible Role of Enzymatic

Degradation of Para-Aramid Fibrils, (1997) 41 Ann. Occup. Hyg., pp. 148-153.135Bernstein, D.M., Graph on Biopersistence of p-Aramid Fibres.136Bernstein, D.M., Summary of the Final Reports on the Chrysotile Biopersistence Study, Geneva,

1998, document submitted to the Panel by Brazil as a Third Party (see above Section IV).137Rowlands, N., Gibbs, G.W. and McDonald, A.D., Asbestos Fibres in the Lungs of Chrysotile Miners

and Millers − A Preliminary Report, (1982) 26 Ann. Occup. Hyg., pp. 411-415.138Oberdörster, G., Macrophage-Associated Responses to Chrysotile, (1994), 38 Ann. Occup. Hyg.,

pp. 601-615.139Bernstein, D.M., Summary of the Final Reports on the Chrysotile Biopersistence Study, Geneva,

1998, document submitted to the Panel by Brazil as a Third Party (see above Section IV).

WT/DS135/RPage 367

the question that must be asked is: what is the risk for workers manufacturing or using the product?The decision on which fibre is safer has to be made on the basis of an assessment of the risk of diseasefor workers when manufacturing and using a product containing chrysotile compared to that whenmanufacturing and using the same product containing the substitute when subjected to the same orequivalent handling.

5.559 Three sources of data might be considered: experimental animal studies; human(epidemiological) studies; and in vitro studies. The latter (in vitro) are of little value for estimatingrisk as they only involve tests of, for example, biological activity in cells isolated from the processeswhich occur in a complete organism. Thus, they are an inadequate basis of comparison to the effectsof inhalation on animals, much less humans.

(ii) Animal Studies

5.560 The first approach involves the exposure of animals to fibres of well-defined characteristicsand concentrations by inhalation and following them for their lifetimes. Such studies have been donefor a wide range of synthetic mineral fibres. Problems with this approach are many, as has beendemonstrated in the considerable work done in recent years on synthetic mineral fibres. First, theanimal species may have a limit on the size of fibre that it can inspire. Second, there are markeddifferences in the sensitivity of different animal species. For example, a refractory ceramic fibre(RCF) which produced one or two mesotheliomas in rats, produced mesotheliomas in 40 per cent ofthe hamsters exposed. Third, the lifetime of rats is about two years. In order to produce an effectwithin the lifetime of the animals they are subject to enormous exposures. Such exposures canproduce the abnormal situation of lung overload so that the real reason for any biological effect is notclear. Fourth, an animal must produce an effect within two years (before it dies of natural causes). Ifbiopersistence is important, fibres that are readily removed in humans over the course of a human'slife are not removed from the lung of an experimental animal because of the high exposures andshorter life. Fifth, the interpretation of much of the experimental work must be done with caution,because until recently, fibre exposures were reported on mass not number basis. As the materialstested can have quite different dimensions, the same mass can lead to exposures involvingconsiderably different concentrations of fibres on a number basis.

(iii) PVA

5.561 There are no studies relating the long-term effects of exposure to PVA fibres.

(iv) Cellulose

5.562 Studies that have been done with cellulose have shown that it initiates a severe inflammatoryresponse140 and fibrosis.141 Unfortunately, no chronic exposure data have been published.

(v) Glass Fibres

5.563 While there have been many studies of glass fibres, the only study in which the samemethodology was applied to the study of synthetic mineral fibres and chrysotile asbestos is that ofHesterberg.142 He found that while there is an increased risk of lung cancer identified at high

140Hadley, J.G., Kotin, P. and Bernstein, D.M., Subacute (28 Day) Repeated Dose Inhalation ofCellulose Building Insulation in the Rat, (1992) The Toxicologist, 225 (abstract).

141Muhle, H., Ernst, H. and Bellman, B., Investigation of the Durability of Cellulose Fibres in the RatLungs, (1997) 41 Ann. Occup. Hyg., pp. 184-188.

142Hesterberg, T.W., Miller, W.C., Theveney, Ph. and Anderson, R., Comparative Inhalation Studies ofMan-Made Vitreous Fibres: Characterization of Fibres in the Exposure Aerosol and Lungs, (1995) 39 Ann.Occup. Hyg., pp. 637-653. Hesterberg exposed rats to a concentration of 10,000 WHO fibres/ml of chrysotile,

WT/DS135/RPage 368

concentrations (as for glass and other fibres) the animal data suggest that at low level exposure, therisk associated with the chrysotile exposure is considerably less than that associated with the syntheticmineral fibres tested.

(vi) Para-aramid Fibres

5.564 While in recent years the information base concerning aramid fibres has increased greatly,there remain several issues. The only data are those derived from experimental studies in animals.While studies of biopersistence suggest that long fibres are shortened by enzymes in the lungs ofanimal experiments (Searl) and hence removed from the lung, the situation in humans is not known.Two researchers (Davis143 and Pott144) have produced mesotheliomas by intra-peritoneal injection ofthese fibres, so their potential to produce mesotheliomas cannot be dismissed. The interpretation of"proliferative keratin cysts" observed during inhalation experiments remains unclear.145 Minty et al., ina criteria document for an occupational exposure limit (OEL) in the UK, summarized what was knownabout the para-aramid fibres at that time and drew several parallels with chrysotile. For example theystate that "[T]he balance of evidence suggests that respirable aramid fibres possess a low potential toproduce mesothelioma which is likely to be at least as low as with chrysotile.146

5.565 Referring to chrysotile, they conclude that mesothelioma "would only be detectable followingvery heavy and prolonged exposures." The recent evidence that the mesothelioma risk for chrysotileminers and millers is associated with tremolite, will render the threshold of mesothelioma fordownstream workers even more remote. These authors considered a clear no-effect level of 2.5 f/mlfor pulmonary toxicity and a recommended OEL of 0.5 f/ml to allow for "uncertainties in interspeciesdifferences."

(vii) Epidemiological Data

(i) PVA

5.566 Drs. Musk, de Klerk, Henderson and Infante did not identify any epidemiological studies ofPVA fibre workers. In fact, there is one study involving a small number of PVA fibre productionworkers (about 400 exposed employees).147 Even though the length of exposure thus far is quite short,already two lung cancer deaths have occurred in the cohort to date. Clearly a much longer follow-upis needed. Regarding mesothelioma, it must be noted that with such a small population, even if halfwere dead and there were one mesothelioma, the risk would be 0.5 per cent which is more than therisk of mesothelioma found in Quebec miners and millers exposed to tremolite contaminated

which resulted in 18.9 per cent lung tumours. Rats exposed to 232 f/ml of one type of glass fibre resulted in5.9 per cent lung tumours, with 4.4 per cent lung tumours reported with other man-made vitreous fibres and13 per cent with a RCF sample. The air control resulted in 1-3 per cent tumours (At 1000 f/ml, the risk of lungtumours would be just under 2 per cent which is well within the rate of tumours in the control animals).

143Davis, J.M.G., Carcinogenicity of Kevlar Aramid Pulp Following Intraperitoneal Injection into Rats,(1987) Technical Memorandum No. TM/87/12 Published by the Institute of Occupational Medicine, Edinburgh,Scotland.

144Pott, F., Roller, M., Ziem, U., Reiffer, F.J., Bellman, B., Rosenbruch, M. and Huth, F.,Carcinogenicity Studies on Natural and Man-Made Fibres with Intraperitoneal Tests in Rats, (1989) In: Non-Occupational Exposure to Mineral Fibres. J. Bignon, J. Peto, K. Saracci eds. IARC Scientific Publication No. 90Publ. International Agency for Research on Cancer, Lyon, pp. 173-179.

145IARC International Agency for Research on Cancer (1997), Monographs on the Evaluation ofCarcinogenic Risks to Humans. Vol. 68.

146Minty, C.A., Meldrum, M., Phillips, A.M. and Ogden, T.L., P-aramid Respirable Fibres CriteriaDocuments for an Occupational Exposure Limit, HMSO (1995).

147Morinaga, K., Nakamura, K., Koyama, N. and Kishimoto, T., A Retrospective Cohort Study of MaleWorkers Exposed to PVA Fibres, (1999) 37 J. Industr. Health, pp. 18-21.

WT/DS135/RPage 369

chrysotile. (Also, the Panel should note that there were no mesotheliomas among 1267 deaths inchrysotile exposed friction product manufacturing workers exposed to chrysotile.) Thus, this studycannot detect either mesothelioma or lung cancer risks as low as at that already known for chrysotile.Clearly, there are no human data on which to assess risk to conclude that the risk is less than it is forchrysotile either per f/ml of exposure or globally from work with products manufactured using PVAfibres.

(ii) Cellulose

5.567 Dr. Infante states that there are three studies in which cellulose exposures have beeninvestigated, but he does not identify them. The other two experts do not suggest any epidemiologicaldata. Studies in which there is no overall increase in mortality from lung cancer are not adequate toinvestigate the risk of exposure. To assess this risk, the relationship between lung cancer andcellulose fibre exposures on a per fibre basis must be, but has not been, examined.

(iii) Para-aramid Fibres

5.568 None of the experts reported any epidemiological data. Clearly, para-aramid fibres can beinhaled, as experimental animals inhaled them. However, because para-aramid fibres have been usedfor such a short time, there are no data on the relationship between levels of fibre exposure and therisks of lung cancer, mesothelioma or other adverse effects for persons working with this substitute orproducts manufactured using it.

(iv) Glass Fibres

5.569 There have been several studies of workers exposed to glass fibres during fibre manufacture.Studies have also included rock [stone] and slag wool exposures. The latter were associated with anincreased risk of lung cancer even at very low levels of exposure. Doll148 concluded that the risksfrom such exposures were greater than those associated with chrysotile asbestos. Doll summarized thesituation as follows: "an occupational hazard of lung cancer has been demonstrated in the rock andslag wool section of the industry and possibly the glass wool section." The human evidence since thattime has not dispelled concern about the risks associated with these fibres. This question is still notresolved.

5.570 Dr. Infante149 and co-workers once reached the same conclusion for glass fibre (although hehas changed his opinion in his current report). In his report, Dr. Infante mentions that after speakingwith workers, he now thinks that there was asbestos exposure at the plant studied by Shannon inOntario, Canada, where a high level of risk of lung cancer was found in glass fibre workers. A recentdiscussion with Dr. Harry Shannon about his study of glass fibre workers reveals that to hisrecollection, no one had raised the question of asbestos as a potential confounder in his study. Henoted that, as the study was published many years ago, it seems unlikely that this issue – if it actuallyexisted – would not have been raised and studied, especially by the glass fibre industry.150 Clearly, nonew analyses have been done, so the impact of supposed asbestos exposure, if it took place, is notknown. Dr. Infante's reversal of opinion does not seem justified, as no new data are presented. Forexample, it is not known whether the "asbestos exposed workers" had high or low glass fibreexposure. If they had low glass fibre exposure, then the risk associated with the glass fibre exposures

148Doll, R., Mineral Fibres in the Non-Occupational Environment: Concluding Remarks, in Bignon, J.,

Peto, J. and Saracci, R., eds., Non-Occupational Exposure to Mineral Fibres, IARC Scientific PublicationNo. 90, 1989, pp. 511-518.

149Infante, P.F. et al., Fibrous Glass and Cancer, (1994) 26 Am. J. Industrl. Med., pp. 559-584.150Gibbs, G., Phone Communication.

WT/DS135/RPage 370

might increase. Thus, without additional analyses, the best estimates at present are the originalanalyses by Shannon.151

5.571 It was noted earlier that exposure levels during production may not be the same as thoseduring product use. While it has not been possible to find data on the use of glass fibres in chrysotilecement or friction products, there has been an estimate of the risk associated with glass fibres asinstalled in homes. In this study, Wilson152 used animal data to derive lung cancer risk estimates forglass fibre exposure. They assumed an exposure of 1f/ml for one year based on available data andestimated that the lung cancer risk in smokers associated with blown glass wool without binder in asmoker without a respirator would be 2.4 x 10-4. If one uses the same methodology as applied bythem to derive a chrysotile estimate (based on epidemiological data), but for friction productmanufacture, the risk would be very much lower: 0.12 x 0.00058 = 0.00007 or 7 x 10-5. This is alower risk than calculated for glass fibres. In fact, there is no demonstrated increased risk of lungcancer in the friction industry, so even this chrysotile risk is hypothetical and certainly anoverestimate. Wilson acknowledges this in their paper.

5.572 In this light, it is safer to work with chrysotile in friction products than to work with glassfibres. While it might be argued that there has been no report of an increased risk of mesothelioma inhumans as a result of manufacturing glass fibres, in the case of chrysotile, there is greater confidenceconcerning this lack of risk because there is no evidence of an increased risk of mesotheliomaassociated with friction products throughout their lifecycle, and the studies are far more voluminousand varied in approach. There are no systematically gathered data available concerning downstreamrisks for glass fibres as used as a substitute in cement or friction products. Similarly, with regard tothe asbestos cement industry, Harrison153 reports that most studies have not found an increase inmesothelioma; certainly this is true for chrysotile asbestos cement plants. Thus, it is evident thatthere are clear no epidemiological or experimental data to conclude that "glass fibres" are safer thanchrysotile, indeed, there is evidence to suggest the contrary.

5.573 In summary, the experts have based their opinions on very limited, if any, data. The data thatdo exist suggest that the conclusions of the Panel's experts concerning the relative safety of thesubstitutes and chrysotile at low concentrations are incorrect.

2. The European Communities

(i) Introduction

5.574 Each of the four scientific experts appointed by the Panel has recently responded to the pointswhich the Panel wished to clarify. The European Communities note that the four experts consultedunanimously and unambiguously corroborate the analysis that led France to adopt the Decree 96-1133banning asbestos. This analysis was communicated to the Panel in the two written submissions of theEuropean Communities of 21 May and 30 June 1999 and is based on the following points:

(a) All forms of asbestos, including chrysotile asbestos, are carcinogens, and there is noscientifically established threshold below which exposure to asbestos would bewithout risk for humans;

151Shannon, H.S. et al., Mortality Experience of Ontario Glass Fibre Workers − Extended Follow-Up,

(1987) 31 Ann. Occup. Hyg., pp. 657-662.152Wilson, R., Langer, A.M. and Nolan, R.P., A Risk Assessment for Exposure to Glass Wool, 30

Regulatory Toxicology and Pharmacology, pp. 96-109.153Harrison, T.W., Levy, W.S., Patrick, G., Pigott, G.H. and Smith, L.L., Comparative Hazards of

Chrysotile Asbestos and its Substitutes: A European Perspective, (1999), Environmental HealthPerspective, 107.

WT/DS135/RPage 371

(b) exposure to asbestos, including chrysotile asbestos, is the cause of many cancers, thevast majority of which affect secondary users, particularly workers coming intocontact with materials containing asbestos, including asbestos cement;

(c) so-called "controlled" use of asbestos is in fact impossible in practice;

(d) there are asbestos substitutes which are far less dangerous for human health.

5.575 In this document, the European Communities do not wish to make systematic and detailedcomments on all the replies by the four experts consulted, but will simply refer to the mainconclusions and give a summary of their replies in the annex.154

(ii) The four experts consulted agree that all types of asbestos, including chrysotile, arecarcinogens and that there is no established threshold under which exposure to asbestos is withoutrisk for humans

5.576 The four scientific experts unanimously consider that chrysotile asbestos, as well asamphiboles, can cause mesothelioma and lung cancer inter alia.

5.577 The four experts also unanimously agree that there is no scientifically established thresholdbelow which exposure would not pose any risk of cancer for humans. All the experts state that therisk of cancer is proportional to the cumulative level of exposure and all consider that thenon-threshold linear model is the most scientifically appropriate model for guaranteeing the level ofhealth protection decided upon by France in this particular case. This explains and confirms that thenon-threshold linear model has always been used, without exception, by the authorities in all thosecountries that have so far carried out scientific assessment of the cancer risk.

(iii) The four experts consider that exposure to asbestos, including chrysotile asbestos, is the causeof many cancers that mainly affect secondary users, particularly workers in contact with materialscontaining asbestos, including asbestos cement

5.578 The four experts consider that the vast majority of the risks concern so-called "secondary"users155, in other words, workers making interventions (building workers, electricians, plumbers,maintenance workers, handymen, etc.) because of their large number and the nature of their activities,even if the individual risks are sometimes lower.

5.579 For example, most cases of mesothelioma now affect this category of workers in all theindustrialized countries, including Canada (Quebec) and Australia, countries which produce asbestos.The four experts point out that the levels of exposure in the course of occasional contacts withasbestos cement products are very high, much higher than the levels at which a risk of cancer has beendefinitely and scientifically established.

(iv) The four experts consider that the so-called "controlled" use of asbestos is not practicallypossible

5.580 The four experts unanimously agree that so-called "controlled" use aimed at ensuring aconstantly low level of release of the fibres into the atmosphere is absolutely impracticable in the vastmajority of work situations where workers have to deal with friable or non-friable materialscontaining asbestos.

154See Annex V to this Report.155 See factual arguments by the EC, Section III.A.4.

WT/DS135/RPage 372

5.581 The four experts consider that it might be possible in very special situations where a smallnumber of workers carry out a very precise task. They also indicate that interventions on materialssuch as asbestos cement can release very large quantities of asbestos fibres; that protective equipmentis not or not always effective and not always used; that the recommended procedures are rarely orincorrectly followed in small enterprises such as those in the building sector; that it is quiteimpossible to apply them to non-professionals (for example, handymen, etc.).

E. SUPPLEMENTARY REMARKS FROM DR. HENDERSON156

1. Concerning the comments from the European Communities

5.582 The comments from the European Communities are very brief, occupying only four pages inEnglish translation, so that only a short comment is needed. The tabular summary of the four experts'reports appears to represent a fair précis of my conclusions and opinions, if an oversimplification. Inpara. 5.579, the European response refers to Canada (Quebec) and Australia as countries that produceasbestos. As pointed out in para 5.27, Australia is no longer an asbestos producer.

2. Concerning the comments from Canada

5.583 At 62 pages and with over 50 Annexes, the comments from Canada are far lengthier than theresponse from the European Communities; the Canadian documents include new information,necessitating more extensive discussion. Some general comments follow; other issues are discussedlater under specific sub-headings.

5.584 In para. 5.441, the comment is made that some of the answers from the experts "appear not todistinguish between chrysotile and amphibole exposure" or between "modern uses ... and historicaluses ... ." Throughout my own Report, I tried to make this distinction wherever appropriate, and myanswers to the Panel's questions deal almost exclusively with chrysotile (like EHC 203 [1]) -e.g. mydiscussion of the risks to brake mechanics157 and the tabulation of risk estimates for lung cancer andmesothelioma (Tables 12 and 13 in paras. 5.203 and 5.205. At the same time, it seems worthreiterating that commercial chrysotile from Canada on average contains variable trace amounts (about<1 per cent) of tremolite (fibrous tremolite is a non-commercial amphibole; e.g. please see EHC 203).In relation to Canada's concerns about the "experts' conclusions on tradesmen" (e.g. buildingconstruction workers), my perspective seems to concur with the IPCS/WHO monograph on chrysotile(EHC 203):

"... (c) Some asbestos-containing products pose particular concern and chrysotile use in thesecircumstances is not recommended These uses include friable products with high exposure potential.Construction materials are of particular concern for several reasons. The construction industryworkforce is large and measures to control asbestos are difficult to institute. In-place building materialsmay also pose risk to those carrying out alterations, maintenance and demolition" ... [p 144].

5.585 My recognition that chrysotile is substantially less potent than the amphiboles on a fibre-for-fibre basis for mesothelioma induction - and that present exposures overall are substantially less thanin the past - explains why my Report dwelt mainly on workplace exposures (e.g. to building materialsand friction products). Non-workplace exposures (e.g. non-occupational exposures includinghousehold contact or neighbourhood exposures [2-4]) - and exposures to friable insulation materials -received less attention and were included mainly to put the present situation into historicalperspective.

156For complete references of the documents quoted in this Section, see Annex III to this Panel Report.157For decades, brake blocks and brake linings used in Australia have contained Canadian chrysotile

asbestos only, with no added amphiboles.

WT/DS135/RPage 373

5.586 In para. 5.489, the Canadian comments state that: "... He [Henderson] believes amphibolesmay be greater than 60 times more likely than chrysotile to induce mesothelioma ... ." In fact, inpara. 5.103, I had stated that:

"There is general though not universal agreement of a differential potency between the amphibolesversus chrysotile for mesothelioma induction; in this respect, the amphiboles are substantially morepotent, with estimates ranging from 2-4X, to 10X, to 12X on a fibre-for-fibre basis, to 30X, to a 30-60Xgreater potency, or more (e.g. please see EHC 203)."

5.587 Later in my Report (paras. 5.141 and 5.413), I gave my estimate that chrysotile has apotency 1/10th-1/30th the carcinogenicity of crocidolite for the mesothelium. This estimate has notchanged.

5.588 Some clarification of my answer to Question 1(b)158 from the Panel seems necessary, takinginto account Canada's quotation of my view and the comment that: "Canada takes note that ... [this is]... overwhelmingly a workplace issue, and therefore not related to the 'handyman'." In my answer, Itook "workplace" to refer to any situation where work of any type is carried out (e.g. cutting, sawing,drilling, grinding, rasping or sanding of asbestos-cement building products), whereas I interpreted theexpression "a larger part of the population" to refer to general environmental exposure to asbestos(e.g. simple occupancy of buildings, or potential exposure of urban dwellers in general to asbestosderived from the brakes of passing vehicles). Obviously, any risks to handymen who carry outmaintenance on homes only occasionally will be much less than the risks to professional tradesmensuch as carpenters, who work day in, day out at building construction sites - because the frequencyand duration of exposures for the handyman will be less (with lower cumulative exposures), assumingthe types of asbestos to be the same. However, this may not always be so. For example, I know thatsome "handymen" in Australia specialize in buying and living in dilapidated houses, to carry outextensive renovations on these dwellings (e.g. throughout each weekend or more often) before sellingthem a year or more later; because the house qualifies as a principal place of residence, the profit isnot taxable. The handyman then repeats this exercise on another "handyman special" house, and soon. Many such handymen also regularly carry out maintenance and renovation work on other homes,so that their exposures may approach those of professional tradesmen, but they style themselves as"home handymen". The activities of such handymen are virtually unregulated because they areself-employed and a number work on a "strictly cash" basis.

5.589 In commenting on the experts' responses to Questions 1(b), Canada reiterates the propositionthat "chrysotile is readily removed from the lung", and estimates of the half-life of chrysotile are givenas 90-110 days, and even a shorter estimate of < 10 days from Dr. David Bernstein. Canada goes onto state that "size for size, chrysotile has a very short half-life." I again draw attention to the 1999study by Finkelstein and Dufresne [5], who found a lung tissue half-life of eight years for chrysotilefibres > 10 µm in length; this investigation was discussed briefly in the Endnote to my report (seeSection V.C.4):

" ... in the past, the kinetics of chrysotile clearance from lung tissue have been investigated mainly inexperimental models using rodents. In an autopsy study published in 1999, Finkelstein and Dufresne[5] ... investigated clearance of chrysotile from the lung tissue of 72 Quebec chrysotile miners andmillers in comparison to 49 control subjects, using regression analyses, with the following findings:

There was a significant association between the duration of occupational exposure and the tissueburdens of chrysotile and tremolite.

The concentration of chrysotile decreased with time after exposure ceased but the concentration oftremolite did not.

158Please see also my answers to Questions 1(e) and 5(a).

WT/DS135/RPage 374

The clearance rate varied inversely with the length of chrysotile fibres. For fibres > 10 µm in length -i.e. fibre lengths in the reported range for carcinogenicity - the clearance half-time was estimated to beeight years. In other words, the tissue bio-persistence of chrysotile fibres in this study seemssubstantially more prolonged than in rodent experiments, and presumably corresponds to persistenthigh chrysotile fibre concentrations for many years after cessation of occupational exposure in humans,as discussed on p 31. It is also notable that the concentration of 6,250,000 chrysotile fibres mentionedon p 31 (for an individual but by no means unusual patient) is probably above the level at which Rogerset al. [6] identified an odds ratio for mesothelioma of > 8.5 (even allowing for differences in fibre sizein the counts by the two different laboratories), and even the duration of 16 years after exposurestopped (as opposed to its commencement: 24 years) falls into the lag-time range for lung cancerinduction by asbestos.

Studies like this suggest that clearance mechanisms can be overwhelmed and break down atoccupational levels of exposure in humans, with the existence of a long-term sequestered fraction ofchrysotile fibres."

5.590 This study seems to be of particular significance for the tissue bio-persistence of chrysotilefibres in comparison to substitute materials (please see below, paras. 5.642 to 5.652).

5.591 I also emphasize that some of the estimates given in my Report were conservative, withpotential under-estimation of effects. For example, after discussing the incidence rate for spontaneousmesothelioma unrelated to asbestos as being in the range of 1-2 mesotheliomas per million person-years - whereas the likely true figure is probably less than one [4] - I nonetheless used the upper figureof two cases/million for comparison with mesothelioma incidence in some occupational groups(e.g. the incidence of mesothelioma among male automobile/brake mechanics in Australia; please seepara. 5.253). In a similar way, I referred to a 30-fold differential rate for lung cancer among theSouth Carolina (Charleston) chrysotile textile workers in comparison to the Quebec chrysotile minersand millers, whereas others give the differential as up to "about 50 times higher in Charleston" [7].

5.592 I also draw attention to the occurrence of mesothelioma among various cohorts and studiesother than the Quebec chrysotile miners/millers, as set out in paragraphs 5.124-5.141, and to theincidence of mesothelioma among mechanics in Australia as shown in the 1999 Report for theAustralian Mesothelioma Register [AMR 99] and in NICNAS 99 (see my answer to Question 2).

5.593 In my Report, I discussed the limitations or deficiencies of those studies which reported anincreased risk of lung cancer only among workers with pre-existing asbestosis (e.g. the Hughes-Weillstudy [8]) - and the uncertainties of the study by Camus et al. [9] on lung cancer risk from non-occupational exposure to chrysotile among females in Quebec159 - but the comments from Canada(para. 5.529) reiterate the Hughes-Weill conclusion that an increased risk of lung cancer occurs "onlyin those with asbestosis." (Please see paras. 5.73-5.74 and 5.152-5.162 above; this subject wasreviewed extensively by Henderson et al. [13] in 1997. Lung cancer risk among the South Carolinachrysotile textile workers versus the Quebec chrysotile miners/millers is also discussed in paras. 5.596to 5.620 below).

159Canada's comments also refer to the meta-analysis of lung cancer risk reported by Lash et al., [10],

which identified a low risk. Meta-analysis is a field that lies outside my expertise, but I understand that there arevarious models for meta-analysis and problems with this approach (e.g. see Blettner et al. [11] who state that "...Meta-analyses from published data are in general insufficient to calculate a pooled estimate since publishedestimates are based on heterogeneous populations, different study designs and mainly different statistical models[Abstract] ... Meta-analyses using published data are, therefore, restricted and seldom useful to produce a validquantitative estimate or to investigate exposure relations such as dose-response [p 8] ..."). In a meta-analysisof 69 asbestos-exposed occupational cohorts, Goodman et al. [12] identified "... meta-SMRs of 163 and 148 [forlung cancer] with and without latency, with significant heterogeneity of results ..."

WT/DS135/RPage 375

5.594 At various points (paragraphs 5.475, 5.498 and 5.545), the comments from Canada quotede Klerk and Armstrong [14], in a chapter on The Epidemiology of Asbestos and Mesothelioma, in thebook Malignant Mesothelioma, for which I was the senior editor and a co-author. I shall leave it toDr. de Klerk to respond.

5.595 In passing, I point out that Malignant Mesothelioma was published in 1992; the text for thosechapters which I wrote was current up to September 1990, and the manuscript was sent to thepublisher shortly thereafter. Much new information on asbestos-related diseases has accumulatedsince that time (e.g. references 15, 16, 113, 125, 126, 131-133, 140, 141, 170-172, 177-179, 181,185-187, and 190-194 in my Report, to list but a few). My views on many aspects of asbestos-relateddisorders have changed very substantially since Malignant Mesothelioma was published (e.g. myviews on asbestos and lung cancer - please see references [13, 15-18] in these SupplementaryRemarks).

(a) Lung cancer rate among South Carolina (Charleston) chrysotile textile workers versus theQuebec chrysotile miners/millers

5.596 With respect to this question, Canada states (see paras. 5.485-5.486):

"Dr Henderson states that the "greater carcinogenicity of the amphiboles [...] appears not to extend tothe induction of lung cancer [p 40], but he admits that 'chrysotile is implicated in one of the lowest ratesof asbestos-associated lung cancer (in Quebec chrysotile miners and millers)' [where I also stated thatchrysotile is also implicated in the highest lung cancer rate]. Dr Henderson's reluctance to conclude thegreater carcinogenicity of amphiboles seems to be caused by the results of Dr Dement's study of theCharleston, South Carolina asbestos textile industry [...]".

"The Charleston data has [sic] recently been revisited by Bruce Case, André Dufresne, A.D. McDonald,J.C. McDonald and Patrick Sébastien in a study released in Maastricht in October 1999 at the VIIthInternational Symposium on Inhaled Particles, a symposium attended by some of the world's leadingexperts. This study shows that a significant amount of crocidolite and amosite fibres was found in thetextile workers' lungs. This analysis sheds new light on the issue and explains the extreme results of theoriginal study by Dr Dement [...] and the subsequent study by Dr Stayner [...]. These studies of textileworkers exposed to crocidolite and amosite can thereby no longer be used to demonstrate the risksassociated with chrysotile fibres."

5.597 Subsequently, the manuscript for a paper by Case et al. [19] entitled Asbestos Fibre Type andLength in Lungs of Chrysotile Textile and Production Workers: A Preliminary Report arrived byfacsimile transmission. I offer the following comments on this document (and, later, on the Abstractfor the corresponding presentation at the Maastricht meeting [20]):

5.598 A disclaimer beneath the title [19] indicates that this is a "DRAFT DOCUMENT: SUBJECT TOREVISION - NOT TO BE CITED". It is cited nonetheless. There is no indication that this document hasgone through a process of peer review and been accepted for publication.

5.599 This study revisits the study reported in 1989 by Sébastien et al. [7], and the draft manuscriptindicates that the same grids were examined (but fewer cases). The main difference between thisinvestigation and the earlier study by Sébastien et al. [7] is that Case et al. [19, 20] analysed longfibres > 18 µm in length, whereas Sébastien et al. [7] studied fibres > 5 µm in length, with an aspectratio > 3:1. (It is common practice for fibre burden analyses to focus on fibres ≥ 5 µm in length andthere is no evidence that the carcinogenicity of asbestos fibres - in terms of lung cancer induction - isrestricted only to fibres about 20 µm in length or more.)

5.600 Another study on the lung fibre content of the Charleston chrysotile textile workers wasreported in 1997 by Green et al. [21]; this investigation studied all fibres resolvable by electron

WT/DS135/RPage 376

microscopy and with an aspect ratio > 3:1. For this study, lung tissue was analysed from 39 textileworkers versus 31 comparable controls matched closely for age (median age at death for the asbestosworkers was 56.0 years, versus 59.0 years for the controls).

5.601 In the Green et al. [21] study, the Charleston chrysotile workers had a higher lung content ofchrysotile in comparison to the controls (geometric mean = 33,450,000 versus 6,710,000 f/g drylung), with a higher content of tremolite (3,560,000 vs. 260,000); the asbestos workers also had aslightly elevated mean amosite/crocidolite content of 470,000 fibres vs. 210,000 for the controls(please see Table 1).

TABLE 1: MINERAL FIBRE CONTENT OF LUNG TISSUE, SOUTH CAROLINA ASBESTOSTEXTILE WORKERS VS. CONTROLS (ALL COUNTS = FIBRES X 106 / G DRY LUNG)*

Textile workers ControlsAge at death (median; years) 56.0 (M); 57.0 (F) 59.0 (M); 62.5 (F)Year of death (median) 1971 (M and F) 1972 (M); 1971 (F)

Chrysotile (fibres x 106/ g dry lung) 33.45 6.71Tremolite 3.56 0.26Amosite/crocidolite 0.47 0.21Anthophyllite 0.16 0.13

Mullite 1.63 4.01Other 1.02 1.9All fibres 52.46 16.02

*Modified from Tables 1 and 3 in Green et al. [21]; M = men; F= females.

5.602 In the discussion section, Green et al. [21] commented that:

"The population was exposed almost exclusively to chrysotile asbestos from Quebec. The native orecontained about 1% tremolite asbestos. The high concentrations of chrysotile and tremolite asbestosfound in the lungs of the asbestos textile workers are also consistent with their exposure histories. Ourfinding on enrichment of tremolite relative to chrysotile in the lungs of asbestos workers is consistentwith previous reports. The presence of crocidolite in some of the lungs of the asbestos workers is inkeeping with the use of small quantities of crocidolite between 1950 and 1975, but the values were onlyslightly greater than those found in the control population. ... The increased risk of lung cancer in theasbestos textile workers is also unlikely to be due to differences in exposure to tremolite asbestos, asSebastien et al. have shown that the textile workers had less tremolite asbestos in their lungs thanminers and millers of the original ore after matching for exposure intensity. Differences in exposure toother commercial amphiboles (crocidolite and amosite) may have played a small part based on our owndata ... and on the data of Sebastien et al., which showed a small excess of these amphiboles in thelungs of the textile workers compared with the miners; however, it is very unlikely that this is thewhole explanation as commercial amphiboles formed a very small proportion of the total amphiboles inboth studies. Moreover, review of the 10 cases with lung cancer in this study on whom lung fibreanalyses were made, showed only one case with substantially increased (> 1 x 106 fibre/g dry lung)crocidolite or amosite".

5.603 In this study, it is also notable that the lung cancer cases on which fibre burden analysis wascarried out were not representative of the cohort as a whole: e.g. autopsies were carried out on onlyabout 10 per cent of all deaths in the cohort, and the mean lifetime cumulative exposure for the tenlung cancer cases was 94.6 fibre-years in comparison to 67 fibre-years for male lung cancer casesacross the whole cohort [21, 22].

WT/DS135/RPage 377

5.604 There are even greater concerns about the representativeness of the cases on which fibreburden analysis was carried out by Sébastien et al. [7]. For example, this study was confined to tissuefrom 72 autopsies among 857 deaths (8.4 per cent) among the Charleston cohort, and there were onlyseven lung cancer cases out of 66, whereas Case et al. [19] list 126 lung cancers, so that the fibreburden data reported by Case et al. [19] appear to deal with no more than 5.56 per cent of theCharleston lung cancers. It is also notable that the mean age at death in the Charleston group wasabout a decade younger than the age at death for the Thetford group which formed the basis forcomparison in the 1989 study by Sébastien et al. [7]160

5.605 In addition, as reported by Sébastien et al. (see Table 3 in reference [7]), those cases from theThetford group that came to autopsy showed an over-representation of asbestos-related diseases (lungcancer, mesothelioma and pneumoconiosis) than the Thetford cohort overall - so that cases of lungcancer + mesothelioma + pneumoconiosis added up to 37 out of 89 autopsies (42 per cent), incomparison to 306 out of 4463 deaths across the whole cohort (7 per cent) [7]. For the Charlestoncohort, the figures were more comparable, so that lung cancer + mesothelioma + pneumoconiosiscases added up to 13 out of 72 autopsies (18 per cent) in comparison to 10 per cent across thecohort [7].

5.606 In the more recent study from Case et al. [19], there is a further point on which the two studygroups (Thetford versus Charleston) are not comparable: the time following cessation of exposurewas a median of eight years for the Thetford group, in comparison to 20 years for the Charlestoncohort (please see Table 2). Therefore, it is clear that those lung cancer cases on which fibre burdenanalysis was carried out from each cohort were not representative of each cohort, and that there werealso substantial differences between the two cohorts for the same types of case. Finally, themanuscript from Case et al. [19] does not include a control group against which the two cohorts can becompared (the only one of the three investigations that does is the Green study [21]).

TABLE 2: COMPARISON OF THE SOUTH CAROLINA AND QUEBEC CHRYSOTILE WORKER COHORTS*

South Carolina textile workers Quebec miners/millersCohort number 3022 10,918Cohort deaths 1258 8009Age at death (years) 67 ± 10 (??)** 56 ± 6 (??)**Lung cancers in cohort 126 [SMR 197] 657 [SMR 137]Mesotheliomas in cohort 2 38Years since cessation of exposure(median)

20 8

Geometric mean exposure (mpcfy)*** 3.63 186Subjects studied 64 43Lung cancer cases studied ?

(7/72 autopsies in ref [7])?

("random" selection of 43 casesfrom 89 original cases that

included 22 lung cancer cases -ref [7])

Chrysotile (fibres x 106/ g dry lung) 0.054 0.231Tremolite 0.027 0.325

160There is a discrepancy between the ages at death in the original paper by Sébastien et al. [7]

(i.e. mean = 55.8 ± 9.7 for the Charleston cohort vs 67.5 ± 9.7 for the Thetford group) and the follow-up studyby Case et al. [19] (Table 1A, where the ages are reversed: 67 ± 10 for the Charleston group vs 56 ± 6 forThetford). Clearly, one or the other must be wrong.

WT/DS135/RPage 378

South Carolina textile workers Quebec miners/millersAmosite/crocidolite 0.037 0.024Total amphibole (tremolite +amosite/crocidolite)

0.064 0.349

*Modified from Case et al. [19]. Fibre counts represent geometric means; all expressed as fibres x 106/ g drylung; **see footnote 160; ***mpcfy = millions of particles per cubic foot-years.

5.607 From the above Table, it is evident that the amosite/crocidolite content of lung tissue from thetextile workers is slightly (< 2-fold) higher than the amosite/crocidolite in the lung tissue from theQuebec miners and millers (37,000 fibres > 18 µm in length versus 24,000). This difference inconcentration seems to be insufficient to explain the "huge" [19] risk difference (about 30-fold) in theslope of the lung cancer dose-response line between the two groups. In addition, it is noteworthy thatthe tremolite content of lung tissue was higher in the Quebec miners and millers than the Charlestontextile workers (325,000 versus 27,000 for fibres with a mean fibre length of 21.7 versus 21.9 µm).The point is that the total amphibole content (tremolite + amosite + crocidolite) is higher in theQuebec miners and millers at 349,000 f/g dry lung in comparison to a total amphibole contentof 64,000 among the Charleston textile workers. In this respect, there is no evidence that tremolite issubstantially less potent than the other amphiboles for lung cancer induction, as shown by the highlung cancer incidence (SMR = 285) among Montana vermiculite miners exposed only totremolite/actinolite (please see paragraph 5.107-5.111).

5.608 From these studies, it appears that the amosite/crocidolite content of lung tissue among theCharleston textile workers may in part be a reflection of low level exposure to the small amount ofcrocidolite (< 1000 kg total) used in the plant from 1950-1975 to make an asbestos tape or braidedpacking. The material was received at the plant as a yarn ready for weaving, and no fibre preparation,carding, spinning or twisting was done using crocidolite. Packing workers were not included amongthe textile worker cohort, and analysis of lung cancer risk by operation in the plant shows alloperations to be at about the same lung cancer risk after controlling for chrysotile exposure in alogistic model (Dement, personal communication, 1999).

5.609 A portion of the amosite/crocidolite content may also be explicable by general environmental(non-occupational) exposure, taking into account the small differences between the amphibole contentin the textile workers versus the controls in the study reported by Green et al. [21]. In this respect,amphibole concentrations of up to 100,000-200,000 fibres per gram (f/g) dry lung tissue can beexpected for about 5 per cent of the population in Germany [23]. Therefore, it seems that theamosite/crocidolite cannot explain the risk of lung cancer in the Charleston cohort in comparison toeither matched controls (also matched for smoking) versus the Thetford miners and millers.

5.610 If major significance is to be assigned to the small difference in amosite/crocidolite content oflung tissue between the Charleston workers versus the Thetford miners/millers for lung cancerinduction, a question that immediately arises is: where are the mesotheliomas among the Charlestonworkers? Case et al. [19] suggest that misclassification of mesotheliomas as lung cancers among theCharleston workers could have produced under-estimation of the true number of mesotheliomas"while having virtually no effect on the lung cancer excess or lung cancer exposure-disease slope ofrisk." No evidence in support of this proposition is adduced, and Case et al. [19] state that this is"speculation." The larger number of mesotheliomas in the Quebec cohort may be explicable in part bythe higher mean total amphibole content for this group, but this still leaves unexplained thedisproportionately larger numbers of lung cancers in the Charleston group (e.g. the ratio of lungcancers to mesotheliomas in the Thetford group is 657/38 = about 17:1, whereas the ratio for theCharleston group is 126/2 = 63:1).

WT/DS135/RPage 379

5.611 Case et al. [19] are also rather more cautious in their interpretation than the propositions putforward in Canada's responses to the reports from the experts. For example, on the last page of textthey state:

"... comparison of groups of individuals using this technique is valid only insofar as those studied arerepresentative of the larger groups ... from which they are derived. We cannot be certain to whatdegree our groups of chrysotile miners/millers and textile workers are representative of the cohortsfrom which they are derived161 ... the two groups are not directly comparable in some ways: not onlywas exposure much higher in the miners/millers, but the interval between cessation of employment anddeath was shorter .... . Our results closely parallel those reported by Sebastien et al.. Any other resultwould be surprising since the subjects were drawn from the latter study. ... Caution remains ininterpretation. ... One continuing mystery, given the apparent non-trivial long-fibre commercialamphibole exposures is the low level of reported mesotheliomas in this cohort ...".

5.612 Given the data on fibre lengths across the cohorts, in comparison to the data in Sébastien et al.[7], the difference in lung cancer rates between the two groups cannot be explained by differences infibre length. This is stated explicitly by Case et al. [19].

5.613 However, on looking at the data, it seems that the differences between the two cohorts mightbe explicable in part by the exposure estimates. Differences in exposure assessment are not refuted bythe "new" study reported in draft form by Case et al. [19] or by the earlier study reported by Sébastienet al. [7]: e.g. the difference between 20 years (Charleston) and eight years (Quebec) for clearanceafter exposure ceased could have a large effect. One can calculate the final exposure (end ofexposure) (N0) from the final fibre content in lung tissue at death (N), from the equation

N/N0 = e-λt

where λ represents a clearance coefficient (λ = 0.693 ÷ T1/2 ) and t = half-life in tissue (T1/2). For T1/2 =8 years [5], λ = 0.693/8, so that for the chrysotile miners/millers, where N = 0.231, N0 = 0.462. For theCharleston textile workers, where N = 0.054, N0 = 0.306.

5.614 If T1/2 is shorter (e.g. one year), then N0 for the miners/millers = 59.2 and the correspondingN0 for the textile workers = 56456.

5.615 Therefore, for a half-life of eight years, one would expect the ratios of exposure (exposureminers/millers ÷ exposure textile workers) to be 0.462/0.306 = 1.5. For a half-life of one year theratio becomes (exposure miners/millers ÷ exposure textile workers) 59.2/56465 = 0.001. (For tissuehalf-lives of 90-110 days or < 10 days, the differences would be even more drastic.) However, theratio of the estimated exposures (mpcfyQuebec/mpcfyCharleston) is 186/3.63 = 50, suggesting that one orother particle count estimate is incorrect.

5.616 In this respect, it might be argued that the exposure estimates for the Charleston cohortrepresented an under-estimation of exposure, but this suggestion is not supported by the low tremolitecontent in the lung tissue of the Charleston workers, and is explicitly rejected by Sébastien et al. [7],who state (p. 187):

"The hypothesis of a systematic underestimation of exposures to asbestos in Charleston, which wouldhave accounted for the difference in risk, must therefore be rejected and other explanations sought."

5.617 Given that contamination of the Charleston chrysotile by mineral oils has now been excluded,one possibility that remains is over-estimation of the exposures for the Quebec chrysotile

161Clearly, from the data in Table 2 and the discussion in paras. 5.604 to 5.609, they are not

representative.

WT/DS135/RPage 380

miners/millers (with under-estimation of risk). If this explanation is unsustainable, it follows that theparadox remains, it remains unexplained, and seems likely to remain so.

5.618 Finally, I draw to the attention of the Panel the following comment by Case and Dufresne [20]in the Abstract for their presentation at the Maastricht meeting:

"Risk assessment for asbestos exposure is based on lung cancer risk for textile workers, rather thanminers/millers."

5.619 In the draft manuscript, Case et al. [19] state only that:

"... suggestions that the textile worker mortality data [are] suitable for chrysotile risk assessment [forlung cancer] should be re-evaluated ... ."

5.620 Therefore, even if one accepts this proposition for the moment, the claim that theSouth Carolina cohort can "thereby no longer be used to demonstrate the risks associated withchrysotile fibres" goes beyond the data in this study. For the reasons discussed in this section, Iconclude that the data in Sébastien et al. [7] and in Case et al. [19] do not detract from the conclusionsdrawn by myself and other authorities from the investigations carried out the South Carolina cohort byDr. Dement and his colleagues [22, 24].

(b) The question of a threshold for the carcinogenicity of chrysotile (lung cancer andmesothelioma

5.621 On this question, I simply reiterate EHC 203:

"Exposure to chrysotile asbestos poses increased risks for asbestosis, lung cancer and mesothelioma ina dose-dependent manner. No threshold has been identified for carcinogenic risks" [p. 144].

5.622 In the absence of a threshold or an agreed alternative (non-linear) exposure-response model,the linear relationship model is widely employed for risk assessment at low levels of exposure.

5.623 As indicated, the precision or validity of this model is not known at low levels of exposureand, as stated by Dr. de Klerk, the model provides a "conservative estimate". This is the point: in theabsence of direct observational data or credible alternative models, the linear model errs - if it does err- on the side of safety, which is appropriate for risk assessment as a prelude to the formulation ofoccupational health and safety and public health policy. The principle is: if there is doubt, play safe(i.e. first, do no harm; primum non nocere).

5.624 In relation to prudent approaches to occupational and public health policy, The Minerals andMetals Policy of the Government of Canada162 states the following (p 7):

"The precautionary principle is an important factor when the Government needs to make a decision inthe face of scientific uncertainties about cause and effect, and when the potential environmentalconsequences are generally considered to be serious or irreversible. This principle was enunciatedclearly as Principle 15 in the 1992 Rio Declaration on Environment and Development(the Rio Declaration) of the United Nations Conference on Environment and Development (UNCED),to which Canada is a signatory:

'Where there are threats of serious or irreversible damage, lack of full scientific certainty shall not beused as a reason for postponing cost-effective measures to prevent environmental degradation.'

162NRCAN, The Minerals and Metals Policy of the Government of Canada: Partnership for

Sustainable Development, Public Works of Canada, 1996.

WT/DS135/RPage 381

The principle complements science-based approaches for the management of risks. Its use is premisedon the recognition that our scientific understanding of the potential magnitude and consequence ofimpacts on human health and the environment of the production and uses of some minerals and metalsmay be incomplete. While there is a need to work toward closing such gaps in our understanding, thereis also a requirement, where potential impacts are 'serious or irreversible', to consider a cost-effectiveprecautionary approach."

5.625 Later, on page 12, the same Minerals and Metals Policy document states:

"... It is generally accepted that, in some cases, the risks associated with certain products or productuses cannot be properly controlled or managed. Consequently, where such a situation exists, theGovernment [of Canada] will either discontinue or prohibit the specific product or product use."

5.626 Three additional points are worth iteration:

• Exposure to commercial Canadian chrysotile is not "chrysotile-only", but usuallychrysotile + trace tremolite exposure, although evidence indicates that chrysotilewhen uncontaminated by tremolite also has the capacity for the induction of lungcancer and mesothelioma.

• Risk estimates for lung cancer and mesothelioma for low levels of chrysotile exposurewere set out in Tables 12 and 13 (see my response to Question 1(d)).

• As stated in section C.1(f)(viii) and in paragraph 5.95, there are no observational dataon the interactive effects of inhaled commercial chrysotile fibres when these aresuperimposed separately and later upon a pre-existing amphibole ± chrysotile burdenwithin lung tissue (?superimpositional additive or multiplicative carcinogenic effect).In my Report, I emphasized that it has been estimated that up to 15-20 per cent ofmen in industrialized societies may have sustained occupational exposures to asbestos(chrysotile/amphiboles), and Rödelsperger et al. [23] indicate that fibre concentrationsof 100,000-200,000 amphibole f/g dry lung tissue may be expected for about 5 percent of the population in Germany. Rödelsperger et al. [23] have also identified adose-response relationship for mesothelioma induction at these low fibreconcentrations. We do not know what the effect of subsequent chrysotile fibreinhalation superimposed upon an existing amphibole burden of this order might be,but NICNAS 99 states the following (p. 61):

"... multivariate analysis of cases found a dose-response relationship for lung fibre content ofcrocidolite, amosite and chrysotile and the development of mesothelioma. Either a multiplicative oradditive model could be used to fit the relative risk/dose coefficients for the various asbestos types. Aprogressive increase in relative risk with increasing fibre content was reported for all fibres ... ."

5.627 Because the risks of both lung cancer and mesothelioma show a dose-response effect relatedto total cumulative exposure levels, it may be expected that later superimpositional inhalation ofchrysotile ± tremolite fibres would aggravate the overall consequences of a pre-existing asbestosburden (i.e. increase the risk further).

(c) The feasibility in practice of "controlled use" of chrysotile asbestos

5.628 In para. 5.510, Canada identifies the feasibility in practice of "controlled use" of chrysotileasbestos as "one crucial issue, which seems to override all other issues" (i.e. the question of whetherthe application of the "controlled use principle" is feasible and credible at all stages in the life cycle ofchrysotile asbestos).

WT/DS135/RPage 382

5.629 As already indicated (see my reply to Question 5), I agree that this proposition is crucial to thedispute before the WTO. However, for the reasons discussed in my Report, I see no requirement toresile from my perception that - although regulation and control of chrysotile and high-densitychrysotile products may be achievable at some points of the life-cycle (e.g. the manufacture of frictionand high-density products) - "controlled use" of this type is not feasible in reality or in practice atothers (e.g. in building construction and other points of end-use).

5.630 No airborne fibre measurements are available for the overwhelming majority of asbestos-related diseases encountered in my everyday practice, even for exposures throughout the 1970s and insome cases extending into the late 1980s. Among my series of asbestos-associated lung cancers andmesotheliomas, I cannot recollect ever seeing actual airborne dust and fibre measurements at points ofend-use (e.g. at building construction sites or shipyards).

5.631 The concentration of total asbestos fibres in lung tissue from one of my cases of asbestos-associated lung cancer was up to 125,000,000 f/g dry lung (up to 108,000,000 amosite + crocidolitefibres), for a worker who had been employed at a major asbestos-cement manufacturing facility forabout 2-3 years (lag-time = 28 years) [15]. I also note Dr. de Klerk's comment about demolition of anold asbestos-cement factory in Sydney in the latter half of 1999 (probably the same factory) "where noobservable precautions of any kind were being taken"163 (Dr. de Klerk, response to Question 1(a)).

5.632 Other interventions on high-density asbestos-cement materials that can lead to high fibreconcentrations are discussed in my first Report (e.g. Kumagai et al. [25]; my answer to question 1(d);please see also the 1980 report by Rödelsperger et al. [26] on exposure to asbestos-cement dust atbuilding sites, which refers to a daily mean airborne fibre concentration of 0.6 f/ml for fibres > 5µm inlength, and "peak concentrations of more than 100 fibres/ml").

5.633 In para. 5.532, the comments from Canada include the statement that chrysotile "can bepainted without fibre release" (presumably including building products). However, painting of suchproducts can cover warning notices and disguise the true nature of the product, so that workers wholater carry out maintenance or renovation work on the same product - and those who recycle the samematerial - may be unaware of its true nature. In my own series of mesotheliomas, it is not uncommonto encounter cases for which the patient was unaware or unsure that he (or less often she) had workedin the past with an asbestos-containing product.

5.634 In one recent case, the patient worked (1973-1988) at a factory where tins and pails wereproduced. In about 1979 she had worked for several months at a conveyor belt that carried the tinsand pails into a fan-forced oven, which appears in retrospect to have been lined by asbestos-containing insulation. The patient was present when maintenance work on the oven was carried out,and she recalled hot air continually blowing from the oven into her face as she worked on theconveyor. After diagnosis of her mesothelioma and its treatment in the late 1990s by radicalpneuropneumonectomy, an asbestos body and fibre analysis on her lung tissue revealed a count of1640 asbestos bodies per gram dry lung, and a total asbestos fibre count of 34,120,000 f/g dry lung(30,770,000 chrysotile fibres164 + 3,350,000 crocidolite fibres). This was the only history of exposurethat was obtainable on exhaustive questioning.

163The potential for misuse of asbestos-containing materials remains, as shown by some prosecutions

(e.g. please see the UK Health & Safety Executive (HSE) press releases E198:98 and E079:99;http://www.hse.gov.uk/press/e98198.htm and http://www.hse.gov.uk/press/e99079.htm), but cases that comebefore the courts almost certainly represent only a small fraction of the misuses, most passing unnoticed byregulatory agencies.

164Please note the high chrysotile count almost a decade after the patient's employment ended.

WT/DS135/RPage 383

5.635 A similar history was also obtained in another mesothelioma case seen on referral in 1999,where a radio assembly worker had used asbestos-containing cloths used to clean soldering irons,together with a history of about one fibre-hour of exposure to four asbestos-cement building sheetsused for maintenance work on his home; only later did I discover that during his work at the radiofactory, he often entered a walk-in fan-forced oven apparently lined by insulation bricks.

5.636 Again, please see the spread of occupations in AMR 99 attached to my Report; a similarspread of occupations is listed by Hodgson et al. [27] in a 1997 report on mesothelioma mortality inBritain165 - e.g. see Table 1 and Fig 1 in the original reference. In footnote 96 to the comments fromCanada, it is stated that:

"The 'controlled use' approach has been endorsed by the WHO in its 1998 Environmental HealthCriteria 203: Chrysotile Asbestos, p. 144. 'Control measures, including engineering controls and workpractices, should be used in circumstances where the occupational exposure to chrysotile can occur.Data from industries where control technologies have been applied have demonstrated the feasibility ofcontrolling exposure to levels generally below 0.5 fibres/ml. Personal protective equipment can furtherreduce individual exposure where engineering controls and work practices prove insufficient'."

5.637 I interpret this passage from EHC 203 differently when it is taken in the context of thepreceding paragraphs; apart from the heading, the complete text on page 144 of EHC 203 is:

a) Exposure to chrysotile asbestos poses increased risks for asbestosis, lung cancer andmesothelioma in a dose-dependent manner. No threshold has been identified for carcinogenicrisks.

b) Where safer substitute materials for chrysotile are available, they should be considered for use.c) Some asbestos-containing products pose particular concern and chrysotile use in these

circumstances is not recommended. These uses include friable products with high exposurepotential. Construction materials are of particular concern for several reasons. Theconstruction industry workforce is large and measures to control asbestos are difficult toinstitute. In-place building materials may also pose risk to those carrying out alterations,maintenance and demolition. Minerals in place have the potential to deteriorate and createexposures.

d) Control measures, including engineering controls and work practices, should be used incircumstances where occupational exposure to chrysotile can occur. Data from industrieswhere control technologies have been applied have demonstrated the feasibility of controllingexposure to levels generally below 0.5 fibres/ml. Personal protective equipment can furtherreduce individual exposure where engineering controls and work practices proveinsufficient.166

e) Asbestos exposure and cigarette smoking have been shown to interact to increase greatly therisk of lung cancer. Those who have been exposed to asbestos can substantially reduce theirlung cancer risk by avoiding smoking."

5.638 When seen in the context of para. c), I take d) to mean that in those situations where exposureis likely or unavoidable, exposure can be reduced or minimized by certain procedures appropriate tothe circumstances (e.g. engineering controls in manufacture/production or best work practices), butEHC 203 has already identified friable products and building products as materials of "particularconcern" and their use is "not recommended", in part because of difficulties of control in theconstruction industry. I do not see paragraph d) as an endorsement of on-going "controlled use".

5.639 A similar sentiment is expressed in NICNAS 99:

165Hodgson, J.T., Peto, J., Jones, J.R., and Matthews, F.E., Mesothelioma Mortality in Britain: Patternsby Birth Cohort and Occupation, (1997), 41 Ann. Occup. Hyg., 129-133.

166Omitted from my original Report because I did not - and do not - take this to be an endorsement of"controlled use", and also because the figure of up to 0.5 f/ml is up to five times higher than the level of 0.1 f/mlmentioned in Question 5(c) from the WTO Panel.

WT/DS135/RPage 384

"Prudent OHS [occupational health and safety] policy and public health policy favours the eliminationof chrysotile wherever possible and practicable [p 139] ...Best practice must be implemented to minimise occupational and public exposure, and to minimiseenvironmental impact, over the remaining period(s) of use [p 140].A risk reduction strategy using all available and appropriate measures is required to ensure that therisks posed by chrysotile are continually reduced and eliminated wherever possible" [p 140].

5.640 NICNAS 99 also goes on to state (p 140):

"In achieving this it is further recommended that:

a) Specific phase-out periods should be set, with stages (over the shortest possible period of time)to encourage and reflect the availability and suitability of alternatives [to chrysotile].

b) Action is taken in the immediate future to prohibit the replacement of worn non-chrysotileoriginal equipment with chrysotile products, as alternatives are now available.

c) No new uses of chrysotile or chrysotile products should be introduced (i.e., an immediateprohibition on new uses).

d) Occupational health and safety authorities take the lead role in considering thisrecommendation and specific strategies to implement it as worker health is identified as themajor concern."

5.641 As stated in the paper by Jarvholm et al. [28] attached to the Endnote (Section V.C.4) for myreport:

"... The first regulation of asbestos [in Sweden] was introduced in the early 1960s and subjects whostarted their occupational career in the 1960s should have been exposed to lower doses on average, thanthose who started earlier. On the other hand, by the 1960s asbestos was being used more extensively sothe number of people exposed to asbestos may have increased. ... More stringent regulations ofasbestos were introduced in the mid-1970s, which led to the sharp decrease in its use. People who haveonly worked under such conditions were born from 1955 onwards. They have not yet reached asufficient latency time for possible mesotheliomas to have developed so the number of cases [is] few.However, the first indication is that they may have a decreased risk compared with earlier birth cohorts.A more certain conclusion can probably not be drawn for another ten years. Thus, the preventivemeasures of the mid-1970s can probably not be evaluated with reasonable precision until around 2005,30 years later.

The present situation in Sweden, that mortality from mesothelioma due to early use of asbestos is of asimilar size to the total number of fatal occupational accidents, is caused by a situation in which atleast 90% of the asbestos used was chrysotile. However, we have no information about the type ofexposure to asbestos among the cases of mesothelioma - whether they had an exposure to crocidolite oramosite. There is some pressure from the asbestos industry world-wide to change the asbestosregulations to allow the use of chrysotile. To evaluate such an experiment would take at least another30 years. Even if the major cause of mesothelioma in Sweden was from types of asbestos other thanchrysotile, it is difficult to see how the benefits from an increased use of asbestos in Sweden couldoutweigh the uncertainty of the risks. A similar prudent approach would also be appropriate in otherEuropean countries ..."

(d) Are substitute fibres safer than chrysotile?

5.642 In para. 5.539, Canada states:

"Dr. Henderson, for his part, recognises that, as with all fibres, the pathogenicity of substitutes isdefined by the "3 Ds" (dimension, dose, durability). He seems also to understand that, due to the (lackof) historical use of substitutes, we cannot fully know the risks of using them. However, he then seemsto ignore the importance of these facts."

WT/DS135/RPage 385

5.643 My comments on the safety or potential biohazards of substitute fibres were based on thefollowing:

• The dimensions and respirability of substitute fibres. For example, it appears thatsynthetic fibres can be engineered to be either shorter than the lengths of asbestosfibres that have been associated particularly with carcinogenicity, or to bepredominantly non-respirable. In contrast, according to Harrison et al. [29]:

"The intrinsic hazardous properties of chrysotile can never be "engineered out", and the potential forharm will always remain. Prevention of ill-health will thus always rely on the control of exposure,something that history has shown cannot be guaranteed. ... Unlike chrysotile, substitute fibers can oftenbe designed or selected to have particular characteristics."

• Dose: reported airborne fibre concentrations from the manufacture or use ofsubstitute (e.g. synthetic) fibres are low - comparable to or lower than the airbornefibre concentrations produced by the manufacture or subsequent use of chrysotile-containing materials. This being so, my conclusions about the relative safety ofchrysotile versus substitute fibres are based primarily on fibre dimensions (discussedabove) and biopersistence (discussed below).

• Durability (biopersistence): in para. 5.552, Canada states the following:

"It is well known that biopersistence is a key parameter. Indeed, the human evidence for chrysotileindicates that it is likely to be one of the main reasons why chrysotile is less dangerous than theamphiboles in respect to mesothelioma risk. This is clearly recognised by three of the four experts, aswell as by INSERM."

Canada then emphasizes the rapidity of clearance of chrysotile from lung tissue, withreference to a 90-110 day half-life for chrysotile in lung tissue, and an even shorter estimate of< 10 days. Again, I draw attention to the recent study from Finkelstein and Dufresne [5] whoestimated a lung tissue half-life of eight years for chrysotile fibres > 10 µm in length inQuebec chrysotile miners and millers. Accordingly, in my survey of the literature, I placedparticular emphasis on the biopersistence of substitute fibres in comparison to chrysotile.

• The relative potency of substitute fibres or chrysotile fibres to produce pathologicalchanges (e.g. genotoxicity/mutagenicity and the capacity for tumour induction).

5.644 Warheit et al. [30] claim that p-aramid fibres are biodegradable in the lungs of exposed rats,with faster clearance times than long chrysotile fibres, which showed greater biopersistence.

" ... p-aramid is biodegradable in the lungs of exposed rats; in contrast, the clearance of long chrysotilefibres was slow or insignificant, resulting in a pulmonary retention of long chrysotile asbestos fibres.The dimensional changes of asbestos fibres as well as the pulmonary cell labelling data indicate thatchrysotile asbestos fibres may produce greater long-term pulmonary effects when compared to inhaledpara-aramid fibrils" [Abstract].

5.645 In 1993, Hesterberg et al. [31] compared the effects of size-separated respirable fractions offibrous glass (FG) with refractory ceramic fibres (RCF) and chrysotile fibres. They found that:

"Exposure to chrysotile asbestos (10 mg/m3) and to a lesser extent RCF (30 mg/m3) resulted inpulmonary fibrosis as well as mesothelioma and significant increases in lung tumours. FG [fibreglassdesignated MMVF10 and MMVD11] exposure was associated with a non-specific inflammatoryresponse (macrophage response) in the lungs that did not appear to progress after 6-12 months ofexposure. The cellular changes are reversible and are similar to the effects observed after inhalation ofan inert dust. No lung fibrosis was observed in the FG-exposed animals. Further, FG exposure resulted

WT/DS135/RPage 386

in no mesotheliomas and no statistically significant increase in lung tumour incidence when comparedto that of the negative control group. These findings, along with previous inhalation studies, suggestthat respirable fibrous glass does not represent a significant hazard for fibrotic or neoplastic lungdisease in humans" [Abstract].

5.646 In a later (1995) study, Hesterberg et al. [32] found that exposure of rats to crocidolite andchrysotile asbestos and to RCF by inhalation induced pulmonary fibrosis, lung tumours andmesotheliomas (41 per cent of hamsters exposed to RCF developed mesothelioma167); fibreglassesMMVF10 and MMVF11, slagwool (MMVF22) and stonewool (MMVF21) did not produce asignificant increase in lung tumours or mesotheliomas168

5.647 In a further study published in 1998, Hesterberg et al. [33] investigated the biopersistence ofsynthetic vitreous fibres and amosite in the rat lung, together with refractory ceramic fibres (RCF1A).They found that "the very biopersistent fibres were carcinogenic" (amosite, crocidolite, RCF1 and tworelatively durable special application fibreglasses designated MMVF32 and MMVF33), whereas "themore rapidly clearing fibres were not" (including rock [stone] wool designated MMVF21,HT stonewool designated MMVF34, slag wool, and insulation fibreglasses designated MMVF10 andMMVF11).169

5.648 An Annex from Canada170 also includes a 1995 document on p-aramid fibres from the Health& Safety Executive (HSE) in the United Kingdom. In a summary statement (p. 22) the HSEdocument states that:

"The balance of evidence suggests that aramid fibres possess a low potential to produce mesothelioma,which is likely to be at least as low as for chrysotile asbestos. While chrysotile is thought to present ahazard with respect to mesothelioma development, current knowledge indicates that the risks for humanexposure are low, and would only be detectable following very heavy and prolonged exposure. Thus, ifin terms of mesothelioma production, aramid fibres are equally, or less hazardous than chrysotile, it canbe concluded that the risks at occupationally relevant levels of exposure would be extremely low."

5.649 The HSE then set an exposure limit of 2.5 f/ml, but in a subsequent document on Substitutesfor Chrysotile (White) Asbestos, the HSE171 commented that:

"There are many long-established alternatives to chrysotile which do not rely on fibre technology. Forexample, corrugated polyvinylchloride (PVC) and steel sheeting can be used instead of asbestos cementsheets.

167The hamster seems to show a propensity for mesothelioma induction in some circumstances

(e.g. SV40 inoculation) but not others; in some studies (Research and Consulting Company) chrysotile did notinduce mesothelioma or lung in hamsters but in rats it produced pulmonary fibrosis, lung tumours andmesotheliomas, so that the rat has been advocated as the most appropriate model for assessment of the humanrisk from fibre inhalation [32].

168Hesterberg, T.W., Miller, W.C., Thevenez, Ph. and Anderson, R., Chronic Inhalation of Man-madeVitreous Fibres: Characterization of Fibres in the Exposure Aerosol and Lungs, (1995) 39 Ann. Occup. Health,pp. 637-653.

169Wilson et al. [34] estimate that fibreglass is 5-10 times less "risky" than chrysotile and they state that"... no one has found any cancer attributable to the manufacture or installation of glass wool fibers ... ." In theirestimates of lung cancer risk from chrysotile, they use the unit carcinogenicity factor of 0.01 (K; used beforethem by the US EPA), and they calculate an absolute excess lung cancer risk of 1.2 x 10-3 for smokers, and 1.4 x10-4 for non-smokers; for 40 yrs exposure at 1.0 f/ml, these estimates equate to 4.8 x 10-2 (smokers) and 5.6x10-3 (non-smokers) - i.e. about 5 per cent and 0.5 per cent respectively, both of which can be considered quite"high". (Wilson, R., Langer, A.M. and Nolan, R.O., A Risk Assessment for Exposure to Glass Wool, (1999) 30Regulatory Toxicology and Pharmacology, pp. 96-109.

170Minty, C.A., Meldrum M., Phillips, A.M., and Ogden, T.L., P-aramid Respirable Fibres CriteriaDocuments for an Occupational Exposure Limit, HMSO (1995).

171HSE document: http://www.hse.gov.uk/pubns/misc155.htm.

WT/DS135/RPage 387

Several types of non-asbestos fibres can also be substituted for asbestos; they have been developed foruse in a wide range of products. The main non-asbestos fibres in current use are polyvinyl alcohol(PVA), aramid and cellulose. A considered scientific view on their safety has recently becomeavailable. In July 1998, the UK's Department of Health Committee on Carcinogenicity (CoC)concluded that these three asbestos substitutes (PVA, cellulose and aramid) are safer than chrysotile.This view was endorsed by the European Commission Scientific Committee on Toxicity, Ecotoxicityand the Environment in September 1998."

5.650 More recently, a press release172 from the UK Health and Safety Commission (HSC/HSE)announced a prohibition on the importation, supply or use of chrysotile in Great Britain, effectivefrom 24 November 1999.

5.651 I also re-emphasize the comments in the reviews quoted in my original Report (answer toQuestion 6), including the review by Harrison et al. [29] who comment along the following lines:

"The diameter of PVA [polyvinyl alcohol] fibres, as manufactured, is well above the respirable limitand most of them are not inhalable. ... the fibres are mostly in the range of 10-16 µm diameter. Thereis evidence that they do not fibrillate (split lengthwise). Many of the particles seen in the atmosphereare non-fibrous. ... Although the published toxicologic information on PVA is relatively sparse, theparent material has been used extensively in surgery and has food contact clearance, presumably basedon unpublished studies. Indications of an accumulation of oligomers in the kidney in somecircumstance173 ... mean that the spectrum of molecular weight of material in the fibres as used shouldbe considered, especially if a smaller diameter material were to be produced. The material will degradeonly slowly, if at all, in the lungs. ... Thus, substitution of PVA for asbestos fibers in products such asasbestos-cement should result in reduced exposures. This prediction has been confirmed in industrialapplications where very low fibre counts have been experienced. Misuse of installed material wouldnot result in significant exposure.

... On balance, the use of aramid fibers should result in reduced levels of fiber exposure as compared tochrysotile asbestos and the fibrils released will be no more toxic and will be less biopersistent. Thepredicted reduction in absolute exposure levels has been achieved in industrial practice. Misuse ofinstalled material would not be expected to give significant exposures.

... On balance, the coarse fiber structure and the long experience in use indicate that substitution ofcellulose fiber for chrysotile asbestos should result in reduced occupational exposures to fiber andlower levels of deposition in the lung. The apparent biopersistence of cellulose in the lung would be apossible cause for concern if the potential for limited lung damage is confirmed.

... We believe that the continued use of chrysotile in asbestos-cement products is not justifiable in theface of available and technically and adequate substitutes. Likewise, there seems to be no justificationfor the continued residual use of chrysotile in friction materials."

5.652 These comments also coincide with one of the recommendations in NICNAS 99:

"... Current overseas experience with the phasing out of chrysotile products indicates that a range ofalternatives is available to suit the majority of uses. Good OHS practice dictates that use of chrysotileshould be restricted to those uses where suitable substitutes are not available, and alternatives shouldcontinue to be sought for remaining uses" [p 139].

172HSC press Release C054:99: http://www.hse.gov.uk/press/c99054.htm.173Referred to in the comments from Canada.

WT/DS135/RPage 388

(e) Summary

5.653 It is my perception that the conclusions in my Report submitted already to the WTO concurwith mainstream thinking and approaches to occupational and public health policy from national andinternational health authorities; these include, inter alia:

• The National Occupational Health & Safety Commission in Australia (WorkSafeAustralia). (Please see NICNAS 99.)

• The World Health Organization (EHC 203).

• INSERM (France).

• The National Health & Safety Commission/Health & Safety Executive (HSC/HSE) inGreat Britain.

• Medical Research Council (MRC) Institute for Environment and Health at theUniversity of Leicester (UK).

• National health authorities in other European Nations.

• The Collegium Ramazzini.

5.654 This being so, it is my perception that the dispute before the WTO is, to some extent, focusedupon inappropriate issues. There has been on-going argument among scientists on the health hazardsof chrysotile asbestos (the chrysotilophiles versus the chrysotilophobes). Given the extent andcomplexity of the scientific literature - with contradictory observations on some important issues andwith uncertainties related to gaps in observational data - it is almost inconceivable that thiscontroversy can be resolved by the WTO Panel, or, indeed, that it will be resolved in the foreseeablefuture (partly because no control group free from asbestos exposure can be assembled to ascertain thetrue spontaneous mesothelioma rate).

5.655 The point to be emphasized is that there exists a substantial body of independent scientific andmedical opinion - embodied in national and international health authorities - that chrysotile iscarcinogenic with no delineated threshold; that it cannot be controlled at all points of end use; andthat existing scientific evidence indicates that safer substitute materials are available.

5.656 To me, this body of opinion is no tendentious artifice designed only to secure a commercialadvantage. From my perspective, this is perhaps the crucial issue, from the so-called precautionaryprinciple, given that neither side is likely to concede that the other has proven its case at a high orderof scientific probability. In other words, the question is not so much whether there exists a provenhealth risk or virtually no risk from the continued use of chrysotile, but whether there exists a body ofindependent and reputable opinion that the possible risks or uncertainties about risk justify a policy ofhighly restricted use or non-use.

5.657 From this perspective, restriction of chrysotile to only a very few special applications - or itsprohibition - is a reasonable and defensible measure designed as a cautious and prudent approach topublic and occupational health policy.

5.658 Therefore, I re-affirm the conclusions set out in my original Report (paragraph 5.431) thatchrysotile should either:

WT/DS135/RPage 389

(a) Be restricted to only a few and well-defined applications174 so that it is inaccessible tothe great majority of workers and is available for use by only small and cohesivespecialized worker groups that can be trained effectively in its controlled use(e.g. analogous to nuclear fuels); this means that chrysotile should not be used inbuilding products (e.g. high-density fibro-cement materials such as asbestos-cementsheets) or friction products.

OR

(b) It should be made inaccessible to everyone, by prohibition, unless the alternativespose equal or greater hazards and equal or greater problems with control.

5.659 In this latter circumstance, the principle is that minimization of exposure is more certain whenno new chrysotile-containing products are introduced into the workplace or the general environment,so that the total amount of asbestos in-place will diminish over time; the problem then becomesprimarily one of minimization of exposure to existing asbestos products during maintenance, repair,removal, demolition and disposal.

____________________

174In the UK HSC Press Release C054:99 announcing implementation of a policy of prohibition of

chrysotile from 24 November 1999, the following specific uses are allowable until 2001-2005:• The use of compressed asbestos fibre (CAF) in gaskets for use with saturated and superheated

steam, and with certain flammable, toxic and corrosive chemicals until 1 January 2001;• The use of CAF in gaskets for use with chlorine until 1 January 2003;• The use of any sheet which, when in a dry state, has a density greater than 1900 kilograms per

cubic metre and is used at temperatures at or above 500°C until 1 January 2003;• The use of asbestos components in aeroplanes and helicopters where this is crucial for their

safe operation until 1 January 2004;• The use of any product consisting of a mixture of asbestos with a phenol formaldehyde or with

a cresylic formaldehyde resin in vanes for rotary vacuum pumps, vanes for rotarycompressors, any bearing or its housing or for split-face seals used to prevent water leakagefrom hydro-electric power generation turbines or from cooling water pumps in power stationsuntil 1 January 2004;

• The use of asbestos in pre-formed joints made from proofed asbestos cloth for sealing thedoors of steam boilers until 1 January 2004;

The use of asbestos in personal protective clothing when used in very high temperatures (500°C ormore) until 1 January 2005.

WT/DS135/RPage 390

VI. SUBMISSIONS FROM NON-GOVERNMENTAL ORGANIZATIONS

6.1 The Panel received four amicus briefs from the following non-governmental organizations:

- Collegium Ramazzini, dated 7 May 1999- Ban Asbestos Network, dated 22 July 1999- Instituto Mexicano de Fibro-Industrias A.C., dated 26 July 1999- American Federation of Labor and Congress of Industrial Organizations, dated

28 July 1999

6.2 These amicus briefs were transmitted to the parties for their information. In their writtenrebuttals of 30 June 1999, the EC incorporated by reference the submission of the CollegiumRamazzini. In a letter dated 18 August 1999, Canada notified the Panel that, bearing in mind thegeneral nature of the opinions expressed by the non-governmental organizations in those submissions,they would not be useful to the Panel at this advanced stage of the proceedings. Should the Panelnonetheless accept the submissions as amicus briefs, Canada believed that the parties should be giventhe possibility to respond to the factual and legal arguments set out in them. In a letter dated3 November 1999, the EC informed the Panel that it was incorporating by reference the amicus briefsubmitted by the American Federation of Labor and Congress of Industrial Organizations, as thatbody supported the EC's scientific and legal arguments in this dispute. The EC also proposed to thePanel that it reject the submissions from the Ban Asbestos Network and the Insituto Mexicano deFibro-Industrias A.C., as those documents contained no information of relevance to the dispute. In aletter dated 10 November 1999, Canada again urged the Panel to reject the four amicus briefs as it wasinappropriate to admit them at this stage in the proceedings. Should the Panel nevertheless considerthese submissions, Canada considered that, for the sake of procedural fairness, the parties should havean opportunity to comment on their content.

6.3 In a letter dated 12 November 1999, the Panel informed the parties that, in the light of theEC's decision to incorporate into its own submissions the amicus briefs submitted by the CollegiumRamazzini and the American Federation of Labor and Congress of Industrial Organizations, the Panelwould consider these two documents on the same basis as the other documents furnished by the EC inthis dispute. It was also on that basis that the Panel submitted those two submissions to the scientificexperts for their information. At the second substantive meeting of the Panel with the parties, thePanel gave Canada the opportunity to reply, in writing or orally, to the arguments set forth in thesetwo amicus briefs. At that same meeting, the Panel also informed the parties that it had decided not totake into consideration the amcius briefs submitted by the Ban Asbestos Network and by the InstitutoMexicano de Fibro-Industrias A.C.

6.4 On 27 June 2000, the Panel received a written brief from the non-governmental organizationONE ("Only Nature Endures") situated in Mumbai, India. The Panel considered that this brief hadbeen submitted at a stage in the procedure when it could no longer be taken into account. It thereforedecided not to accept the request of ONE and informed the organization accordingly. The Paneltransmitted a copy of the documents received from ONE to the parties for information and notifiedthem of the decision it had taken. At the same time, it also informed the parties that the same decisionwould apply to any briefs received from non-governmental organizations between that point and theend of the procedure.

_______________